# DepEd AI Guidelines: What DO 003, s. 2026 Allows and Bans

DO 003, s. 2026 sets DepEd's AI rules: learner use from Grade 4, no personal data in AI tools, declared AI use, banned uses, and duties for school heads.

This is the Markdown copy of [DepEd AI Guidelines: What DO 003, s. 2026 Allows and Bans](https://nash.com.ph/blog/deped-ai-guidelines-do-003-s-2026), an article on the NASH Reviewer Blog by Mark Anthony Llego, published Nov 9, 2026. Topics: School leadership, Aspiring school heads. NASH Reviewer is an independent practice platform not affiliated with DepEd, BHROD, or NEAP.

DepEd's rules on artificial intelligence (AI) are in [DepEd Order No. 003, s. 2026](https://www.deped.gov.ph/2026/02/20/february-20-2026-do-003-s-2026-foundational-guidelines-on-artificial-intelligence-ai-in-basic-education/), the Foundational Guidelines on Artificial Intelligence in Basic Education, signed by Secretary Sonny Angara on February 20, 2026. AI may be used in DepEd schools and offices only as a support tool. Teachers may use it to prepare lessons, materials and assessments; learners in Grades 4 to 12 may use it themselves, as their teachers allow; offices may use it to draft and analyze. People make every final decision, personal information stays out of AI platforms, AI use is declared, and some uses, such as emotion recognition and social scoring, are banned outright.

The order is a framework. It "serves as the overarching framework for all subsequent issuances and initiatives on the responsible and ethical use of AI in basic education", and "Specific operational and technical guidelines shall be issued separately" (Sec. II). Two orders issued on June 4, 2026 already apply it: [DepEd Order No. 016, s. 2026](https://www.deped.gov.ph/2026/06/04/june-4-2026-do-016-s-2026-guidelines-on-lesson-planning-and-learning-design/) to lesson plans and [DepEd Order No. 015, s. 2026](https://www.deped.gov.ph/2026/06/04/june-4-2026-do-015-s-2026-revised-guidelines-on-classroom-assessment-grading-system-and-awards-and-recognition-for-the-k-to-12-basic-education-program/) to classroom assessment, and each has its own guide on this blog ([lesson planning](https://nash.com.ph/blog/deped-lesson-planning-do-016-s-2026), [assessment and grading](https://nash.com.ph/blog/deped-classroom-assessment-do-015-s-2026)). DepEd's AI research center, which builds AI systems that must meet these rules, has its own guide: [ECAIR: DepEd's AI Research Center and Its Projects](https://nash.com.ph/blog/deped-ecair-ai-research-center).

## Key findings

- **Learners use AI tools themselves only from Grade 4.** "Direct learner engagement with AI tools shall be restricted to Key Stages 2 to 4 (Grades 4 to 12)"; in Kindergarten to Grade 3, AI is limited to teacher preparation and differentiated materials (DO 003, s. 2026, Sec. VI.1.1).
- **Learners aged 13 and below use AI only under teacher supervision, after parents are notified** (Sec. VI.1.1), and at the start of each school year teachers tell parents which AI tools they will use, the risks and the safeguards (Sec. VI.2.5).
- **Documents with personally identifiable information may not go into AI platforms** (Sec. VII.1.1), and in data and research work, entering personal information into AI systems "is categorically prohibited" (Sec. VI.1.3).
- **AI uses are sorted into four risk levels.** Emotion recognition, biometric categorization, manipulative chatbots for minors, social scoring and facial recognition scraping are banned; AI in grading, admissions, scholarships, discipline or placement needs a privacy impact assessment, human oversight, audit logging and an appeal mechanism (Sec. VI.1.5, Annex B), and DO 015, s. 2026 bars AI from determining final grades (Table 6).
- **AI detectors "shall not be used as sole evidence of academic dishonesty"** (Sec. VI.1.2(v)), and AI may not be "the primary basis for final judgments on learners' performance and academic honesty" (Sec. VI.1.2(ii)).
- **AI use is declared.** Teachers indicate it in lesson plans and materials, learners cite it in AI-assisted work, and research papers generated by AI "without significant human contribution and proper disclosure" are forbidden (Secs. VI.1.3 and VI.1.7, Annex A).
- **No AI system may be deployed in a school before a privacy impact assessment and an entry in the DepEd AI Registry** (Secs. VII.1.6 and VII.1.7). The school head keeps a record of AI tool use and AI-related incidents (Sec. VIII.4.4).

## What DO 003, s. 2026 covers

The order covers "teaching and learning, assessment, teacher professional development, and education governance and operations across all DepEd governance levels, subject to contextual considerations such as resource availability and technical capacity" (Sec. II). It also sets standards for dealing with AI providers and partners. Public schools and DepEd offices must follow it. "Private schools may adopt these guidelines consistent with their objectives and existing laws and regulations" (Sec. II).

Its rationale is plain about why it was needed: AI use in classrooms and offices "has outpaced the ability of the basic education system to put in place clear, unified, and enforceable policies", and schools and teachers "have relied on informal and inconsistent practices" (Sec. I). The policy statement sets the balance it aims for: "students continue to develop and exercise critical thinking, problem-solving, and other essential cognitive skills while exploring the opportunities that AI offers" (Sec. IV).

The Basic Education AI Framework in Sec. V has three areas, which give the order's general guidelines their structure:

| Area | What it governs | Where in the order |
| --- | --- | --- |
| AI in Education | AI tools that support instruction, personalized learning and assessment | Sec. VI.1 |
| Education on AI | AI literacy, ethical awareness and critical use of AI among learners, teachers and personnel | Sec. VI.2 |
| AI for Education Systems | AI in governance, planning, monitoring and school operations | Sec. VI.3 |

Four principles guide all three, drawn from ASEAN and UNESCO frameworks (Sec. V):

1. **Human-centered and pedagogically appropriate use.** Learners and teachers stay at the center and "maintain control over the AI tools".
2. **Inclusion, equity, and linguistic and cultural diversity.** AI tools "shall remain accessible to the poorest and most disadvantaged learners", with attention to learners in geographically isolated and disadvantaged areas (GIDA), learners with disabilities, and learners with literacy and numeracy problems. AI in Key Stages 2 to 4 builds on the literacy and numeracy of the earlier grades.
3. **Protection of human agency.** AI "shall not compromise the intellectual and relational skills" of learners, teachers and personnel; learners and teachers are told what data AI may collect from them and are kept from becoming dependent on AI. Learning resources such as textbooks, curriculum guides and lesson plans stay "primarily human-created".
4. **Risk-proportionate regulation.** Stricter controls for high-risk uses, room for "safe innovation" with low-risk tools.

The order also places DepEd within the National AI Upskilling Roadmap (NAIUR), which assigns DepEd Levels 0 to 2: foundational digital, literacy and numeracy skills for learners, AI awareness for the general public, and digital and data upskilling for workers and teachers (Sec. V, Table 1). It names three roadblocks to AI in classrooms: "Basic connectivity and infrastructure", "Building teacher capacity", and "Preparing learners" (Sec. V).

**When it took effect.** The order takes effect after its approval and issuance, "15 days after its publication in the Official Gazette or in a newspaper of general circulation and after its publication on the DepEd website", with copies filed with the Office of the National Administrative Register (item 4; Sec. XII.2). DepEd posted it on its website on February 20, 2026, and its later orders treat it as in force: DO 015, s. 2026 "adopts and shall be guided by DO No. 3, s. 2026" (para. 31). Its repealing clause is general: provisions of earlier issuances "inconsistent with these guidelines are repealed, rescinded, or modified accordingly" (item 3). Violations "shall be dealt with administratively" (Sec. IX).

### How the order was announced, and what has followed

| Date | What happened | Source |
| --- | --- | --- |
| February 20, 2026 | Secretary Angara signs DO 003, s. 2026; DepEd posts it on its website | DO 003, s. 2026 |
| February 25, 2026 | Angara explains it to reporters: "It's a graduated policy; the more independent learners will be given greater latitude," and "We just need the students to disclose how they use AI" | Philippine Daily Inquirer, via Asia News Network, February 26, 2026 |
| February 26, 2026 | Backing a new national AI hub, Angara says DepEd will "harness AI to close learning gaps, strengthen governance, and empower teachers", always "with safeguards, transparency, and human oversight" | GMA News, February 26, 2026 |
| March 19, 2026 | The 2026 National Schools Press Conference rules bar AI-generated content as a participant's sole work | DM 021, s. 2026 |
| June 4, 2026 | DO 015 and DO 016, s. 2026 apply DO 003 to assessment and lesson plans. The same day, DepEd and Google announce AI training and certification, with "more than one million teachers and staff" able to use Gemini "through their official education accounts" | DO 015 and DO 016, s. 2026; GMA News, June 4, 2026 |
| August 5, 2026 | Angara says DepEd is preparing a master plan for AI in schools: "The master plan involves all our teachers being conversant with AI." | The Philippine Star, August 5, 2026 |
| August 27, 2026 | Microsoft announces it will give "1 million Filipino public school teachers access to Microsoft Copilot" | Microsoft, August 27, 2026 |

As of November 9, 2026, DepEd has not issued the "Specific operational and technical guidelines" that Sec. II promises; the later rules on AI are those in DO 015 and DO 016, s. 2026, and DM 021, s. 2026 for the press conference. Commentators have asked for more classroom detail. A Philippine Normal University policy brief of June 2026 calls the framework comprehensive but says "it provides limited contextualized guidance for classroom-level implementation" (Javier, 2026).

## Who may use AI, and from what grade

![Illustration of who may use AI tools, by grade, under DO 003, s. 2026. In Key Stage 1, Kindergarten to Grade 3, teachers may use AI only for their own preparation and for differentiated materials, and learners do not use AI tools themselves. From Grade 4 to Grade 12, Key Stages 2 to 4, teachers may use AI in instruction, materials, assessment and feedback as an auxiliary tool whose outputs they check, declare and judge, and learners may use AI tools directly as a support tool, never a substitute for their own effort, citing AI help in what they submit, within limits their teachers set. Two rules follow the learner's age, whatever the grade, and one comes every school year: learners aged 13 and below use AI tools only under strict teacher supervision and after prior parental notification; for learners under 18, parental consent is mandatory and documented before an AI system processes their data; and at the start of each school year teachers tell parents which AI tools they will use, the risks and the safeguards.](https://nash.com.ph/blog/_astro/ai-guidelines-2-who-may-use.D16N6AZp_2i1QRk.webp)

*Figure 1. Who may use AI tools, by grade. Illustration by Mark Anthony Llego, based on DO 003, s. 2026, Secs. VI.1.1, VI.1.7, VI.2.4, VI.2.5, VII.1.6 and VIII.5.*

The grade rule is the first thing a school needs to apply. "Direct learner engagement with AI tools shall be restricted to Key Stages 2 to 4 (Grades 4 to 12). Consequently, AI utilization in Key Stage 1 (Kindergarten to Grade 3) is limited exclusively to teacher preparation and the development of differentiated instructional materials" (Sec. VI.1.1). A Grade 2 teacher may use AI to prepare leveled reading passages, but Grade 2 learners do not use the AI tool themselves.

Two age rules and a yearly notice apply on top of the grade rule:

- **Aged 13 and below.** "access to AI tools shall be strictly supervised by teachers and requires prior parental notification to ensure compliance with child safety standards" (Sec. VI.1.1). The rule follows the learner's age, not the grade, so one class may have learners on both sides of it.
- **Under 18.** When an AI system processes learners' personal data, "Consent must be freely given, informed, and documented prior to processing. Parental consent is mandatory for learners under 18 years old" (Sec. VII.1.6(iii)). This is one of the tests in the privacy impact assessment of any AI system deployed in schools.
- **Every school year.** "At the commencement of each school year, teachers are mandated to inform parents of the specific AI tools to be utilized in instruction, including the associated risks and the ethical safeguards implemented to mitigate them" (Sec. VI.2.5). Parent-Teacher Associations are "the primary platform for regular updates" on AI in the classroom (Sec. VI.2.5).

The order does not define "prior parental notification" further. A note sent home, a school year orientation or a PTA meeting record all show that parents were told before the first use; the safer course is a written notice the school can file.

Phones are a separate rule. [DepEd Order No. 006, s. 2026](https://www.deped.gov.ph/2026/03/24/march-24-2026-do-006-s-2026-guidelines-on-ensuring-a-safe-and-motivating-learning-environment-esmle/) requires schools to "prohibit the use of cellphones and other portable electronic devices by both learners and personnel during instructional periods", with exceptions "in meritorious cases" for academic purposes such as "Accessing educational platforms, digital resources, or participating in authorized online collaborative activities" (Sec. VI.5.g). A class activity that uses an AI tool on learners' phones must therefore fit that exception as well as the AI rules.

## What teachers may do with AI

Teachers "shall exercise professional autonomy in utilizing AI as an auxiliary instructional tool, subject to strict verification of content accuracy to preclude sole reliance on AI-generated outputs" (Sec. VI.1.1). AI is meant to build understanding, "rather than to provide direct answers to assessments" (Sec. VI.1.1). Under that rule, the order lists what teachers "may" do, and two later orders set the details for lesson plans and assessment.

| Task | What is allowed | What is not | Rule |
| --- | --- | --- | --- |
| Instruction | AI to support core competencies and 21st-century skills, "without replacing essential pedagogical practices" | AI in place of the teacher's role | DO 003, Sec. VI.1.1(i) and (iii) |
| Instructional materials | AI in creating and improving materials, "subject to existing quality assurance and approval processes" | Learning resources that are not "primarily human-created" | DO 003, Secs. V (Principle 3) and VI.1.1(ii) |
| Lesson plans | AI for grammar, spelling, clarity, formatting and translation (Guided use), and for rephrasing or organizing text after the teacher's decisions are made and colleagues consulted (Limited use) | AI defining objectives, unpacking competencies, designing learning experiences or judging learner needs (Prohibited use); "fully AI-generated lesson plans" | DO 016, s. 2026, paras. 23 and 25 |
| Writing tests and tasks | AI to draft examinations, quizzes and activity sheets, with "Human supervision, validation, and professional judgment" | Assessment that is "fully automated or solely AI-generated" | DO 003, Sec. VI.1.2(i); DO 015, s. 2026, Table 6 |
| Checking learners' work | AI to help verify learners' outputs | AI as "the primary basis for final judgments on learners' performance and academic honesty"; AI detectors as "sole evidence" | DO 003, Sec. VI.1.2(ii) and (v) |
| Feedback | AI to draft timely feedback that the teacher validates and adapts; the teacher "retains full accountability" | Unvalidated AI feedback | DO 003, Sec. VI.1.2(iv); DO 015, Table 6 |
| Grades | Marking help on multiple-choice or short-answer items, with the teacher checking the scores (Annex B lists this as limited risk) | "Determining FGs" (final grades); "replacing professional judgment"; AI as "a substitute for evaluating learner work" | DO 003, Sec. VIII.5.8, Annex B; DO 015, Table 6 |
| Learner data | Nothing personal goes in | "personal learner data or official DepEd documents" in AI platforms | DO 015, para. 37; DO 003, Sec. VII.1.1 |

Grading needs care because the texts approach it from two sides. DO 003 lists grading among the "areas with significant consequences for learners" that make an AI application high risk (Sec. VI.1.5(ii)), while its Annex B puts "AI grading support (e.g., using AI to auto-mark multiple choice or short-answer items)" under limited risk. DO 003 does not reconcile the two. DO 015, s. 2026 draws the line for classroom assessment: no AI for "Determining FGs" or "evaluating outputs without validation" (Table 6). A tool that marks multiple-choice or short-answer items, with the teacher checking and recording the scores, fits Annex B's limited-risk example. DO 003's own teacher duty points the same way: teachers avoid "the use of AI as a substitute for evaluating learner work, such as grading, marking exams, or providing feedback on performance tasks" (Sec. VIII.5.8).

The same holds for lesson plans. Annex B of DO 003 calls "Lesson plan and test generation" a limited-risk use for teachers, but DO 003 itself keeps lesson plans "primarily human-created" (Sec. V, Principle 3), and DO 016, s. 2026, the later and more specific order, limits what AI may do in a lesson plan and ends: "fully AI-generated lesson plans shall not be allowed" (para. 23). DO 016's own Annex B, the Lesson Planning Guide, has a "Declaration of AI use" row that points to DO 003's Annex A. The [lesson planning guide](https://nash.com.ph/blog/deped-lesson-planning-do-016-s-2026) explains the three categories with examples.

Teachers also carry duties toward their learners' AI use (Sec. VIII.5): to set "boundaries for AI use by clearly communicating to learners when, how, and to what extent AI tools may be used in learning activities and assessments" (VIII.5.2), to balance AI with activities that build basic and critical thinking (VIII.5.3), to monitor how learners use AI tools and report "findings or incidents to the school head" (VIII.5.7), and to teach ethical AI use (VIII.5.6). When appropriate, they also integrate AI topics and ethics "across various learning areas" (Sec. VI.2.1).

## What learners may do with AI, and academic honesty

From Grade 4, learners may use AI tools "to support and enhance their schoolwork", never as "a substitute of their own effort" (Sec. VI.2.4). Teachers make sure learners use AI "as a supporting tool to enhance critical thinking, problem-solving, and creativity, not as a replacement for the development of these skills", and keep giving them independent work that requires their own thinking (Sec. VI.2.4).

The order gives five sample uses, with the purpose of each (Sec. VI.1.7(ii), Table 1):

| Type of use | Description in the order | Examples it names |
| --- | --- | --- |
| Brainstorming | "AI used to generate outlines, ideas, topic suggestions" | ChatGPT, Google Gemini |
| Writing improvement | "AI used to enhance writing mechanics" | Grammarly, Quillbot |
| Presentation creation | "AI used to make slides or scripts" | Canva, Tome |
| Homework support | "AI used for content explanation" | Khanmigo, "ChatGPT with teacher supervision" |
| Teacher feedback enhancement | "AI tools that analyze classroom dialogue to improve instruction" | TeachFX |

The examples are "Illustrative Global Examples", not an approved list, and "teachers shall have the discretion on which types of use are acceptable and which types of use undermine the learning outcomes" (Sec. VI.1.7(ii)).

For assessment, DO 015, s. 2026 turns that discretion into three categories that teachers assign to each task (para. 36, Table 5):

| Category | Tasks | Rule for learners |
| --- | --- | --- |
| Prohibited AI use | Tasks requiring independent recall, reasoning or performance, such as examinations, supervised assessments and in-class writing | No AI; "unauthorized use may constitute academic misconduct, subject to school policies" |
| Limited AI use | Tasks requiring learner thinking and contextual application | AI may support grammar, translation, vocabulary, brainstorming and initial research, "not for generating full outputs"; use disclosed with tools and prompts |
| Guided AI use | Complex tasks requiring higher-order thinking | "Recommended for KS3–KS4"; learners show ownership and reasoning; AI use documented, "including interactions with the tool" |

Term examinations "and other supervised assessments shall be AI-prohibited, except when authorized for accessibility or approved purposes" (DO 015, para. 35). Where AI is permitted, "equivalent non-AI pathways shall be provided", so learners without devices or connectivity are not at a disadvantage (DO 015, para. 38). Written work done at home needs "drafts, notes, proper citation of sources (including AI tools), and in-class validation" (DO 015, para. 33), and performance tasks may end in "demonstrations or oral defense" (DO 015, para. 34). The [classroom assessment guide](https://nash.com.ph/blog/deped-classroom-assessment-do-015-s-2026) covers DO 015 in full.

### AI detectors are not proof

A detector's score may start a conversation; it cannot end one. "AI detectors shall not be used as sole evidence of academic dishonesty. Consequently, teachers shall mandate multi-faceted evaluation strategies—including in-person performance tasks and classroom-based activities—to verify learner authenticity without exclusive reliance on technological verification tools" (Sec. VI.1.2(v)). Teachers must also inform learners "about how AI tools are used in assessments" and provide "opportunities for learners to respond to AI-generated feedback or findings" (Sec. VI.1.2(v)).

> **Caution**
>
> A zero or a dishonesty finding based only on an AI detector's percentage breaks Sec. VI.1.2(v). Before any finding, hear the learner and check the work in class, for example through an oral explanation of the essay or a short supervised writing task on the same topic.

Contests have their own rules. For the 2026 National Schools Press Conference, "artificial intelligence (AI)-generated content cannot be submitted as the participant's sole work", and AI-based image enhancement tools were not allowed in the photojournalism contest area ([DepEd Memorandum No. 021, s. 2026](https://www.deped.gov.ph/2026/03/19/march-19-2026-dm-021-s-2026-2026-national-schools-press-conference/), para. 11 and the photojournalism rules, item 4).

## Declaring AI use

![Illustration of how AI use is declared under DO 003, s. 2026. Teachers indicate AI use in lesson plans, teaching materials, assessments and their results (Secs. VI.1.7(ii), VI.2.2(ii) and VIII.5.5). Learners cite AI help in every AI-assisted submission: the tool, the help it gave, and a confirmation that they understand the work, and teachers check the declarations (Sec. VI.1.7(ii)). In research and data work, every AI tool used to process data is declared on the Annex A form, and no personal or sensitive information goes into AI systems (Sec. VI.1.3). A citation of AI use names the tool, the prompt where it applies, and a link to the chat history (Sec. VI.1.7(i)). When AI use is not disclosed: not disclosing, by itself, is not a punishable offense; repeated or intentional misrepresentation may fall under academic honesty provisions; a learner's undisclosed AI use may be academic dishonesty under school policy; and research papers made by AI without significant human contribution and disclosure are forbidden, with administrative sanctions.](https://nash.com.ph/blog/_astro/ai-guidelines-3-declaring.CgeM1wEW_ZTjbUH.webp)

*Figure 2. Declaring AI use: who declares, what a citation names, and what happens when AI use is hidden. Illustration by Mark Anthony Llego, based on DO 003, s. 2026, Secs. VI.1.3, VI.1.7, VI.2.2 and VIII.5.5, and Annex A.*

The declaration rules sit at three strengths:

1. **Encouraged, for everyone.** "Where appropriate, teaching and non-teaching personnel and learners are encouraged to disclose their use of AI tools in the creation of instructional materials or academic outputs" (Sec. VI.1.7). Schools "may adopt context-appropriate disclosure templates or practices" (Sec. VI.1.7).
2. **Required, for teachers' materials and learners' submissions.** "As applicable, teachers shall explicitly indicate AI use in the preparation of lesson plans, instructional guides, teaching portfolios, teaching materials, learning assessment development and results" (Sec. VI.1.7(ii)). "Teachers shall ensure that all AI-assisted learner submissions include explicit citations detailing: (1) the AI tool employed, (2) the nature of its support, and (3) a confirmation of the learner's independent understanding" (Sec. VI.1.7(ii)). Two later orders make declaration mandatory in their areas: "any use of AI assistance in lesson planning shall be declared" (DO 016, s. 2026, para. 25), and in assessment "Permitted AI use shall be explicitly declared" and "Any use of AI shall be disclosed and documented" (DO 015, s. 2026, paras. 31(c) and 37).
3. **Required, with sanctions, for research and data work.** Personnel "are required to formally declare all AI tools employed in data processing using the prescribed declaration form in Annex A", and AI-generated research submitted "without significant human contribution and proper disclosure is strictly forbidden and shall be subject to applicable administrative sanctions" (Sec. VI.1.3).

A citation of AI use names the tool and, where it applies, the prompt, and "shall include access to the chat history with the AI tool, through a link" (Sec. VI.1.7(i)). The order's example: "OpenAI. (2025). ChatGPT (July 7 version) (link to the chat session or history). Used ChatGPT to generate insight or recommendation regarding the symbolism of the tabo in the book El Filibusterismo" (Sec. VI.1.7(i)). The intent "is to promote honesty and critical reflection, not to impose formal citation rules on every instance of AI use" (Sec. VI.1.7(i)). For research papers, see [how to cite AI tools in APA 7](https://nash.com.ph/blog/cite-deped-order-apa-7).

Annex A has three sample declarations: one for learners, teachers and non-teaching personnel, one for teachers, and one for learners. Each names the tool, the task, the purpose and the prompts used. The learner's version reads in part: "I acknowledge that AI is only a tool that can help support my learning, but it cannot replace my own effort, thinking, and learning." The teacher's version affirms that "no confidential learner information or sensitive institutional information data was shared with the AI platform/app".

What happens when AI use is hidden depends on who hid it and how often:

- "Non-disclosure shall not, by itself, constitute a punishable offense, but repeated or intentional misrepresentation may be subject to applicable academic honesty provisions" (Sec. VI.1.7).
- For learners, AI use "must be treated as a reference source; failure to disclose such use may constitute academic dishonesty" (Sec. VI.1.7(ii)), handled under "school policies on academic dishonesty" (Sec. VI.1.7(ii)).
- Declarations "shall be required as part of academic documentation and shall be retained for review or reference purposes in compliance with the Data Privacy Act of 2012 and records management policies" (Sec. VI.1.7(ii)).

The first two rules lead to the school's own academic honesty policy: teachers guide learners on attribution and responsible AI use "aligned with school-level academic honesty policies" (Sec. VI.1.7). The order does not say what such a policy contains. A school that has none needs one before it can apply these rules evenly, and writing it with teachers, parents and learners is one of the steps for school heads below.

## Teaching learners about AI

The second area of the framework, "Education on AI", is about learners who understand AI, not only use it. DepEd wants learners who "are not only users of AI but are also informed citizens who can critically evaluate AI-generated information, safeguard against misinformation, and participate meaningfully in an AI-driven society" (Sec. VI.2). Where appropriate, teachers integrate "topics on the use and ethical considerations of AI across various learning areas" (Sec. VI.2.1(i)) and make sure AI "promotes and preserves critical thinking, problem-solving, and creativity" (Sec. VI.2.2(i)). Teachers "shall actively demonstrate the critical evaluation and refinement of AI-generated responses" when they use AI for feedback (Sec. VI.1.2(iv)).

The Central Office's Learning Systems strand leads "the integration of AI into the curriculum, including the establishment of implementation timelines" (Sec. VIII.1.4). As of November 9, 2026, no curriculum order has set those timelines. The Strengthened Senior High School curriculum, [DepEd Order No. 017, s. 2026](https://www.deped.gov.ph/2026/06/04/june-4-2026-do-017-s-2026-strengthened-senior-high-school-curriculum/), lists no AI subject among its electives; the nearest are Empowerment Technologies and Fundamentals of Data Analytics (Annex A).

## The four risk levels, and the uses DepEd bans

DO 003, s. 2026 sorts AI by the harm it can do, following the European Union's AI Act of 2024, the ASEAN Expanded Guide on AI Governance and Ethics on generative AI (2025), and UNESCO and UNICEF guidance (Sec. VI.1.5). The level decides what must happen before a tool is used. The order calls this classification "the primary basis for evaluating AI admissibility", and it guides the privacy impact assessment every AI deployment needs (Sec. VI.1.5).

| Level | What it covers | What the order requires | Section |
| --- | --- | --- | --- |
| Unacceptable risk | Uses that "pose significant threats to the rights, safety, or well-being of learners and education stakeholders": emotion recognition, biometric categorization, manipulative chatbots for minors, social scoring, and untargeted facial recognition scraping | Banned. Annex B labels them "Absolutely prohibited" | VI.1.5(i), Annex B |
| High risk | Uses "with significant consequences for learners, such as grading, admissions, scholarships, disciplinary actions, or placement decisions" | Permitted only with "a completed PIA, human oversight ('human-in the-loop'), audit logging, and an appeal mechanism" | VI.1.5(ii) |
| Limited risk | Tools that interact with learners or staff "but do not make determinations with major consequences", such as chatbots for administrative queries and AI-generated text, photos, videos and audio | "transparency obligations, clear disclosure to users, and safeguards against misuse" | VI.1.5(iii) |
| Minimal risk | Spam filters, grammar correction, IT process automation | "standard information technology controls" | VI.1.5(iv) |

![Illustration of DO 003, s. 2026's four levels of AI risk, with sample uses from Annex B. Unacceptable risk, absolutely prohibited: AI that steers which senior high school strand a Grade 10 learner takes; apps that pressure learners with shaming messages; emotion recognition that rates teachers from learners' faces; reliability scores that affect promotions; AI that alters finance, procurement or HR records; cameras rating staff from facial expressions. High risk, permitted subject to strict safeguards (a completed PIA, a human in the loop, audit logging and an appeal mechanism): proctoring by webcam, screen or audio; face recognition to log attendance; AI performance ratings from learners' test scores; AI ranking of applicants for teaching posts; AI shortlisting of job applicants; AI flagging procurement or fund releases as high risk. Limited risk, needing transparency obligations: AI tutoring bots and study companions; adaptive platforms that set lessons or pacing; auto-marking of multiple-choice or short-answer items; bots answering learners' questions on class requirements; drafting office memoranda and reports; dashboards predicting enrollment or budget use. Minimal risk, subject to standard IT controls: spell checkers, text-to-speech and speech-to-text, transcription and captioning, slide layout suggestions, IT helpdesk chatbots, and citation formatting.](https://nash.com.ph/blog/_astro/ai-guidelines-1-risk-tiers.DmyWCZ_S_ZhrroF.webp)

*Figure 3. Four levels of AI risk in DepEd, with sample uses for learners, teaching personnel and non-teaching personnel. Illustration by Mark Anthony Llego, based on DO 003, s. 2026, Sec. VI.1.5 and Annex B.*

Annex B gives sample uses at each level for learners, teaching personnel and non-teaching personnel, and its footnote explains why the same tool can sit at different levels: "AI risk classifications apply across all stakeholders, as they are determined by the system's purpose and context of use rather than the specific user." A chatbot that explains a math problem is limited risk; a chatbot that steers a learner's career is banned. Some examples a school head will meet:

- **Unacceptable for learners:** "Algorithm-based steering of learner's career choices (e.g., using AI to determine or recommend what strand a Grade 10 student should take in SHS.)", chatbots that pressure learners "into studying longer hours with shaming messages", and AI that distorts history, facts or political alignment.
- **High risk for learners:** mental health and wellness apps, which may give information and help identify concerns early but "shall not be used for diagnosis, treatment, or counseling of mental health conditions"; AI proctoring by webcam, screen or audio; scholarship eligibility checkers; and "Biometric attendance/monitoring (e.g., face recognition to log class presence)".
- **High risk for staff:** AI for teacher appraisal, recruitment and promotion, payroll, leave and benefits processing, procurement risk flags, and background checks that cross-check applicants' social media.
- **Unacceptable for staff:** scoring teachers through surveillance alone, ranking teachers by learners' facial expressions, "reliability" or "trustworthiness" scores that affect promotions, and AI that alters finance, procurement or HR records "without accountability".

The ban on emotion recognition has one exception. Health, developmental or assistive analyses using biometric data, "limited to the identification of potential support needs and not involving automated decision-making", are permitted "subject to applicable approval, safeguards, consent, clearance and oversight" (Sec. VI.1.5(i)). The list is not closed: DepEd "shall periodically update the list of permitted and prohibited applications" (Sec. VI.1.5(i)).

Separate from the risk levels, Sec. VI.1.6 lists "inappropriate uses of AI" that teachers, non-teaching personnel and learners "shall avoid":

1. **AI as a substitute for human participation or authority:** automated decision making, AI outputs in place of stakeholder engagement or participatory inputs, negotiations or legal contracts without human validation, and "evaluating the performance of an employee".
2. **AI as an authority or sole source of data:** falsifying data from raw data, official statistics without human validation, speculative data that could pass as actual data, AI as an author or co-author, "writing policies, essays, or speeches solely with AI", and implementation plans without human validation.
3. **AI for emotion recognition, social scoring or manipulative chatbots**, and chatbots used "to generate hateful speech used to harm others".

For a school head, item 2 reaches documents a school produces every year. A School Improvement Plan, an annual implementation plan, a report's statistics or a speech for recognition day may be drafted with AI help, but not written "solely with AI" or released "without human validation".

## Personal data, privacy impact assessments and the AI Registry

The rule a teacher is most likely to break by accident is in Sec. VII.1.1: "internal, confidential, and sensitive documents that contain personally identifiable information shall not be uploaded, shared, or processed through AI platforms, especially those that collect, store, or use input data for model training." A class list, a School Form with learners' names, a learner's anecdotal record, a grievance letter or a teacher's leave papers stay out of chatbots and other AI platforms. For data and research work the rule is absolute: "the input, processing, or storage of personal and sensitive information within AI systems is categorically prohibited" (Sec. VI.1.3). The one opening is for systems DepEd itself governs: processing personal data "within internally governed DepEd AI systems may be permitted solely for approved research or operational purposes, subject to data minimization, anonymization, consent where applicable, and appropriate privacy and security safeguards" (Sec. VII.1.1).

The same section ties AI use to the laws that already protect learners and staff: the Data Privacy Act of 2012 (RA 10173), the Cybercrime Prevention Act of 2012 (RA 10175), the Anti-OSAEC and Anti-CSAEM Act (RA 11930), the Safe Spaces Act (RA 11313), the Intellectual Property Code (RA 8293), "other child protection laws" and DepEd's child protection policy (Sec. VII.1.1). AI systems that process learner data must also meet the National Cybersecurity Plan and the Philippine e-Government Interoperability Framework (Sec. VII.1.2).

Before any AI system reaches a school, four checks apply:

| Check | Who does it | What it covers | Section |
| --- | --- | --- | --- |
| Digital Maturity Assessment | Applied to the school before AI technologies are implemented there | Seven dimensions of readiness (Annex C) | VII.1.4 |
| Contractual warranties from the vendor | Required before any AI is developed or deployed | DepEd data not used to train commercial models; content provenance or watermarking enabled; breaches reported to DepEd within 72 hours. Systems DepEd builds or runs itself for approved purposes are exempt | VII.1.5 |
| Privacy Impact Assessment (PIA) | The initiating office, facilitated by the Data Privacy Office | Scope and purpose, deployment risks, and four tests: transparency, rights of data subjects, consent, security and accountability. "Parental consent is mandatory for learners under 18 years old" | VII.1.6, Annex D |
| DepEd AI Registry entry | The initiating office, with the Education Futures Office (EFO), the Learning Systems strand and the ICT Service | The tool, its vendor, purpose and risk level, PIA results, responsible officers and conformity assessment | VII.1.7, Annex E |

Vendors and partners must also supply "model cards, safety test reports (e.g. bias, toxicity, hallucination rates), content provenance solutions, and agree not to train models on DepEd data" (Sec. VI.3.5(v)). Every partnership on AI needs a memorandum of agreement, a data sharing agreement or a similar agreement covering scope, duration, roles, data handling, feedback and monitoring, and exit or renewal terms (Sec. VI.3.5(i); Sec. X.4). The Strategic Management and Learning Systems strands and the ICT Service "shall review, assess, and evaluate any AI tool or technology proposed for use in DepEd offices and public schools" (Sec. VI.3.5(iii)), and at the division level the Schools Division Office (SDO) must "register and assess all AI initiatives by completing the DepEd PIA and AI Registry prior to any partnership or deployment" (Sec. VIII.3.5).

The assessment is also a duty under the National Privacy Commission's own rules. [NPC Circular No. 2023-06](https://privacy.gov.ph/wp-content/uploads/2024/03/NPC-Circular-Repeal-16-01-Signed.pdf) (December 1, 2023), on the security of personal data, provides that "A PIA should be undertaken for every processing system of a PIC or PIP that involves personal data" (Sec. 5), meaning every personal information controller or processor, and DO 003 requires the PIA to evaluate compliance with it (Sec. III.7.4). The Commission has also applied the Data Privacy Act to AI directly: [NPC Advisory No. 2024-04](https://privacy.gov.ph/wp-content/uploads/2025/02/Advisory-2024.12.19-Guidelines-on-Artificial-Intelligence-w-SGD.pdf) (December 19, 2024) holds the controller "accountable for the processing of AI systems and for the outcomes and consequences of such processing when personal data is involved" (Sec. 2.B), and lists privacy impact assessments and human review of AI output among the governance measures it expects (Sec. 2.B.2). The PIA guide in Annex D is the National Privacy Commission's privacy toolkit form. It asks whether a project collects new or sensitive information about people, maps personal data flows, tests the project against privacy principles, and rates each risk as impact times probability on a 1 to 4 scale each: 1 is negligible, 2 to 4 low, 6 to 9 medium and 10 to 16 high.

The DepEd AI Registry is "a mandatory system maintained by the Education Futures Office (EFO) to record all AI deployments within DepEd and schools" (Sec. III.5), and "the official reference for monitoring, evaluation, and transparency of AI use in basic education" (Sec. VII.1.8). Annex E shows its columns with three entries dated August 29, 2025, each marked "Pilot":

| Entry | Risk classification | Intended use, as printed | PIA | Conformity assessment |
| --- | --- | --- | --- | --- |
| ChatGPT (OpenAI) | "High Risk (education context with minors; misinformation & privacy risks)" | "K–12 classroom support (teacher-facilitated), teacher preparation, administrative drafting; not for automated grading or decisions." | Passed | "Not yet assessed" |
| Microsoft Copilot | "Limited to High Risk (depending on data integration and PII handling)" | "Administrative drafting, report generation, teacher workload reduction (document prep). Not for automated student assessment." | Passed | "Pending evaluation" |
| Google Gemini (Google DeepMind) | "High Risk (content generation, privacy, and bias concerns)" | "Teacher support, translation, and drafting of learning resources. Restricted in use for direct student evaluation." | Passed | "Pending evaluation" |

Each row names a proposed focal office as its responsible officer ("To be assigned") and a human-in-the-loop safeguard, such as "Teacher reviews and approves outputs; fact-checking required; prohibit entry of student PII". The rows show how DepEd classed three general-purpose tools in August 2025; the order does not approve or ban any brand. Whatever tool a teacher uses, the rules on grade levels, personal data and declaration still apply, and a "teacher-facilitated" entry does not open AI tools to Kindergarten to Grade 3 learners.

Gemini is already open to teachers through DepEd accounts, and Copilot has been promised. More than one million teachers and staff can use Gemini "through their official education accounts" (GMA News, June 4, 2026), and Microsoft has announced it will give "1 million Filipino public school teachers access to Microsoft Copilot" (Microsoft, August 27, 2026). DO 003 makes no exception for tools provided this way. A teacher using Gemini or Copilot on a DepEd account still keeps documents with personally identifiable information out of it (Sec. VII.1.1), keeps personal learner data and official DepEd documents out of assessment work (DO 015, s. 2026, para. 37), checks its output, and declares its use.

## The Digital Maturity Assessment Tool

Annex C rates a school on seven dimensions, each on five levels: Starting (1), Developing (2), Established (3), Evolving (4) and Leading (5). The school records its self-assessment and remarks for each dimension.

| Dimension | Starting (1) | Leading (5) |
| --- | --- | --- |
| Pedagogy | Mostly teacher-directed, lecture-based, "one-size-fits-all" lessons | Student-centered, personalized learning "fully embedded in the daily classroom" |
| Assessment | Mostly memory and recall tests; feedback limited to raw scores | Varied online and in-person assessment; learners track their own progress |
| Digital Citizenship | Few people aware of responsible technology use | A school-wide "culture of digital responsibility and safety", backed by active policies |
| Teacher Capacity | Few teachers integrate technology | Teachers innovate on their own and share practices with colleagues |
| Leadership | No school-wide vision or plan for digital learning | Leaders drive a shared vision and allocate resources, with "a clear digital learning strategy" |
| Platform | No platform for digital learning | A school-wide learning management system used by everyone |
| Infrastructure | Unreliable power and connectivity; few people with devices | Infrastructure supports full integration of technology in daily activities |

The tool's Leadership row describes the school head. At Established (3), "School leaders provide consistent support for technology integration through school-level initiatives"; at Evolving (4), they "frequently monitor its progress and support the capacity building of their teachers". The order does not set a minimum level a school must reach before using AI. It requires that schools be "subjected to" the assessment before AI technologies are implemented (Sec. VII.1.4), and it promises priority for schools in geographically isolated and disadvantaged areas (GIDA) in access to digital infrastructure under the DepEd Computerization Program (Sec. VI.2.3(vi)).

## AI in school offices, governance and research

AI may help run a school and an office, within limits:

- **Drafting and idea work.** Personnel may use AI to search for references, brainstorm and develop scenarios, and as support in drafting letters, summaries and policy briefs, outlining presentations and speeches, and checking grammar, but "not used as a substitute for human production and validation" (Sec. VI.3.1).
- **Operations.** Offices and schools "may" use AI for "drafting reports, correspondences, and other administrative tasks (such as creation of presentations, documentation, and schedulers)", and may build dashboards with AI-assisted data visualization, trend analysis and reporting (Sec. VI.3.3).
- **Supervision.** Offices and schools oversee "the proper utilization of AI in processes", make sure AI is used to safeguard personal data, and help review data management practices "to prevent data breaches, hacking, and unauthorized access" (Sec. VI.3.2).
- **Decisions.** In administrative processes, AI "shall not, under any circumstance, make final decisions"; "final decisions shall still be made by DepEd personnel" (Sec. VI.1.4). In policy work it "shall not, under any circumstances, replace human decision-making and stakeholder consultation processes" (Sec. VI.1.4).

Research has the tightest rule. AI in data and research management "is strictly limited to ancillary support functions, such as the analysis of processed data, grammatical refinement, and citation verification" (Sec. VI.1.3). Personnel "are required to formally declare all AI tools employed in data processing using the prescribed declaration form in Annex A", and "the submission of research papers generated, in whole or in part, by AI without significant human contribution and proper disclosure is strictly forbidden and shall be subject to applicable administrative sanctions" (Sec. VI.1.3). The action research guides on this blog apply these rules step by step: [ethics and privacy](https://nash.com.ph/blog/action-research-ethics-deped), [citing AI tools in APA 7](https://nash.com.ph/blog/cite-deped-order-apa-7) and the [review of related literature](https://nash.com.ph/blog/review-of-related-literature-action-research).

## Who does what under DO 003, s. 2026

Sec. VIII assigns duties at every level, seven of them to the school head.

| Level | Main duties | Section |
| --- | --- | --- |
| Central Office | Issues implementation guidelines and builds partnerships; forms a Technical Working Group led by the Strategic Management Strand (EFO), with the ICT Service and Learning Systems, to manage the AI Registry, evaluate AI technologies and issue supplemental advisories; designs AI literacy training through NEAP (teaching personnel) and BHROD (non-teaching personnel); leads AI in the curriculum through the Learning Systems strand; provides quality-assured AI tools and resources | VIII.1 |
| Regional Office | Technical assistance to divisions; uses research on AI in education; monitors and reports regional implementation through the Office of the Regional Director and the Curriculum and Learning Management Division (CLMD) | VIII.2 |
| Schools Division Office | Builds schools' capacity; gives technical assistance to teachers and school heads; oversees data privacy, child protection and ethical standards; monitors through the Office of the SDS and the Curriculum Implementation Division (CID); registers and assesses all AI initiatives through the PIA and the AI Registry before any partnership or deployment | VIII.3 |
| School head | Provides professional development on ethical and effective AI use; orients teachers, learners and parents on data privacy, informed consent and academic dishonesty; consults the community; keeps a monitoring system with records of AI tool use and AI-related incidents; ensures equitable access; manages resources through the Learning Resource Coordinator; ensures compliance | VIII.4 |
| Teacher | Designs AI-enhanced lessons aligned with the curriculum; sets boundaries for learners' AI use; balances AI with activities that build thinking; personalizes instruction; discloses AI use in lesson plans and materials; trains learners on ethical use; monitors classroom use and reports incidents to the school head; does not use AI as a substitute for evaluating learner work | VIII.5 |

At the school level, the order also says "Schools and Learning Centers are mandated to rigorously monitor and document the ethical use of AI by teachers and learners" (Sec. XI.2). The school head's record of AI tool use and incidents (Sec. VIII.4.4) is that documentation, and it feeds the division's reports to the region and the Central Office (Secs. VIII.3.4 and XI.3).

## Training, funding and monitoring

**Training.** DepEd "shall ensure that teaching and non-teaching personnel receive adequate and continuous professional training" on AI (Sec. VI.2.3). Training is phased into three credentialed competency levels:

| Level | What personnel at this level can do | Section |
| --- | --- | --- |
| Basic | Understand AI concepts and ethical use; pick suitable tools; use AI in low-stakes tasks "(e.g., drafting, brainstorming, accessibility support)"; follow ethical and privacy standards | VI.2.3(i) |
| Intermediate | Design AI-enabled learning activities; evaluate AI output; interpret predictive analytics; identify AI bias and stereotypes; guide learners | VI.2.3(i) |
| Advanced | Lead institutional AI strategies; "conduct PIA"; mentor colleagues; work with AI providers; contribute to policy and research | VI.2.3(i) |

Training covers AI functions, teaching uses, assessment and verification, ethics, data privacy, digital responsibility, lesson planning and the development of learning resources with AI (Sec. VI.2.3(ii)). It is delivered through mechanisms schools already run: In-Service Training (INSET), School Learning Action Cell (LAC) sessions, collaborative expertise, coaching and mentoring (Sec. VI.2.3(iii)), and it is to be integrated into the NEAP system, with BHROD and NEAP designing "structured, tiered training pathways, including certification and appropriate incentives" (Sec. VI.2.3(iv)). Personnel are "encouraged", not required, to take part (Sec. VI.2.3(iv)).

**Funding** (Sec. X). Program Support Funds go to every regional office, charged to the EFO's General Management and Supervision funds, under separate guidelines. AI in the curriculum and learning delivery is charged to Basic Education Curriculum funds, and training of teaching personnel on responsible AI use to Human Resource Development funds. Partnerships are formalized by a memorandum of agreement or understanding, and all spending follows accounting, budgeting and auditing rules.

**Monitoring** (Sec. XI). The Strategic Management and Learning Systems strands run a continuous monitoring and evaluation system. Results go to the next higher level, and "non-compliance or identified issues shall be addressed through corrective measures or appropriate administrative, civil, and criminal sanction" (Sec. XI.3). DepEd "shall publish an Annual AI in Basic Education Report detailing registered systems, PIA compliance, capacity-building, incidents, regional equity (including GIDA schools), and mitigation actions" (Sec. XI.4), and will adapt its guidance each year to reports of bodies such as the Independent International Scientific Panel on AI (Sec. XI.5).

**Liability.** "The violation of this provision and or any part of this Order shall be dealt with administratively" (Sec. IX). Administrative cases against DepEd personnel follow DepEd's rules on administrative cases; see [the guide to administrative cases and grievances](https://nash.com.ph/blog/deped-administrative-cases-grievances-guide).

## What this means for school heads

1. **Orient the school early in the year.** The order makes it the school head's duty to "Facilitate advocacy and orientation for teachers, learners, and parents regarding data privacy, informed consent, and the prevention of academic dishonesty" (Sec. VIII.4.2). Pair it with the teachers' duty to tell parents, at the start of the school year, which AI tools they will use, the risks and the safeguards (Sec. VI.2.5). A general PTA assembly can carry both; keep the attendance sheet and the notice as records.
2. **Write the school's rules on AI with the people they affect.** Update the academic honesty policy so it says how learners declare AI help, what counts as misrepresentation, and what happens next, consistent with Sec. VI.1.7 and DO 015, s. 2026's three task categories. Write it with teachers, parents and the community, as indicator 3.8.2 expects of learner discipline policies, listen to learners too, and make sure it never lets a detector score stand as proof (Sec. VI.1.2(v)).
3. **Apply the grade and age rules.** No direct learner use of AI tools in Kindergarten to Grade 3; supervised use and prior parental notification for learners aged 13 and below (Sec. VI.1.1). When the activity runs on learners' phones, it must also fit the academic-purpose exception in DO 006, s. 2026 (Sec. VI.5.g).
4. **Keep personal data out of AI platforms.** Remind the registrar, the guidance designate, the property custodian and every adviser that class lists, School Forms, anecdotal records, health records and personnel files do not go into AI platforms (Sec. VII.1.1), and that in data work nothing personal goes in at all (Sec. VI.1.3). The [school records guide](https://nash.com.ph/blog/deped-school-records-cav-school-forms) explains who may see which records and how they are released.
5. **Route every AI offer through the division.** A vendor, a partner or an LGU offering an AI platform is welcome, but no AI system is deployed in a school before its privacy impact assessment and its AI Registry entry (Secs. VII.1.6 and VII.1.7), and the division registers and assesses every AI initiative "prior to any partnership or deployment" (Sec. VIII.3.5). Ask for the agreement that covers data handling (Sec. VI.3.5(i)) and the vendor's warranties (Sec. VII.1.5). Do not let learners try it before then.
6. **Know the school's digital maturity.** The Digital Maturity Assessment Tool (Annex C) must be applied before AI technologies are implemented in a school (Sec. VII.1.4). Its seven dimensions also make a fair baseline for the technology part of the School Improvement Plan.
7. **Keep a log.** The school head must "Establish a monitoring system with systematic record-keeping to track AI tool usage and any AI-related incidents" (Sec. VIII.4.4). A simple register will do: the tool, the grade and class, the purpose, the date, the parental notice, and any incident (a data leak, a harmful output, a dishonesty case) with the action taken. Teachers report findings and incidents to the school head (Sec. VIII.5.7), and the division consolidates reports (Sec. VIII.3.4).
8. **Train teachers through LAC sessions and INSET.** The order names both as the mechanisms for AI training (Sec. VI.2.3(iii)), and the school head must "Provide professional development opportunities for teachers focused on the ethical and effective application of AI in the classroom" (Sec. VIII.4.1). Start with what teachers already do: AI help with lesson plans under DO 016, test drafting under DO 015, and checking AI output for errors and bias. The [LAC guide](https://nash.com.ph/blog/deped-learning-action-cell) explains how to plan the sessions.
9. **Protect learners without devices.** "Ensure equitable access to AI learning opportunities for all learners, regardless of background or ability" (Sec. VIII.4.5). Where AI is allowed in a task, give an equivalent non-AI pathway (DO 015, s. 2026, para. 38) or school devices in a supervised period.
10. **Use AI in the office, and keep the decisions.** AI may draft reports, letters and presentations and power dashboards (Sec. VI.3.3), but it does not decide (Sec. VI.1.4). It may not rate a teacher's performance on its own (Sec. VI.1.6(i)); Annex B treats AI appraisal of staff as high risk and surveillance-only or emotion-based ratings as banned. Review and own every figure in a report before it is signed.

These steps serve several PPSSH indicators: policy implementation and review ([Domain 1](https://nash.com.ph/blog/ppssh-domain-1-leading-strategically), strand 1.3); records management and emerging opportunities and challenges ([Domain 2](https://nash.com.ph/blog/ppssh-domain-2-managing-school-operations), strands 2.1 and 2.6); learning assessment and learner discipline ([Domain 3](https://nash.com.ph/blog/ppssh-domain-3-teaching-and-learning), strands 3.5 and 3.8); professional development of school personnel ([Domain 4](https://nash.com.ph/blog/ppssh-domain-4-developing-self-and-others), strand 4.5); and communication and community engagement ([Domain 5](https://nash.com.ph/blog/ppssh-domain-5-building-connections), strands 5.4 and 5.5).

## How the NASH may test this

The items below are original practice items, scored 3 (Relational), 2 (Multistructural), 1 (Unistructural) or 0 (Prestructural). An option that breaks a rule scores 0.

### Question 1

*Indicator 2.6.2, Emerging opportunities and challenges. Career Stage 2, Situational, Moderate.*

You are the school head of a central elementary school in a rural town. A technology company offers free accounts on its AI reading tutor for Grades 1 to 6. Learners would read aloud into tablets, and the app would store their names and voice recordings on the company's servers. Your teachers are eager, the parents' association has offered to buy tablets, and about a third of your learners have no device at home. What is the most appropriate action?

- **A.** Welcome the offer, refer it to the division for its privacy impact assessment and AI Registry entry, and plan shared tablets for use from Grade 4.
- **B.** Refer the offer to the division for its privacy impact assessment and AI Registry entry, and plan the tutor for Grades 4 to 6 on devices the learners own.
- **C.** Accept the free accounts for all six grades now, since the tool improves reading, and ask the teachers to collect parents' signed consent as learners log in.
- **D.** Decline the offer for the coming school year, because a third of the learners have no device at home and an AI tutor would only widen the reading gap.

**Answer and explanation**

**Best answer: A.** Points: A = 3, B = 2, C = 0, D = 1.

- **A (3 points, Relational).** It treats the offer as an opportunity, routes it through the division for the privacy impact assessment and registry entry required before deployment, keeps direct use to Grades 4 and up, and plans shared devices so learners without one at home are not left out.
- **B (2 points, Multistructural).** It follows the right channel and the grade limit, but it plans the tutor on learners' own devices, which leaves out the third of the learners who have none.
- **C (0 points, Prestructural).** It puts Grades 1 to 3 on an AI tool, which the guidelines limit to teacher preparation in Kindergarten to Grade 3, and it sends learners' personal data to an outside platform before any assessment. Consent gathered at login does not cure either breach.
- **D (1 point, Unistructural).** It sees the equity problem, but it closes the opportunity instead of managing it; the guidelines ask school heads to ensure equitable access to AI learning, not to refuse it.

**Why.** This aligns with Career Stage 2 expectations because the school head manages an emerging opportunity in a way that keeps it fair for learners who lack devices. Direct learner engagement with AI is restricted to Grades 4 to 12, a privacy impact assessment comes before any AI deployment in schools, and the division registers and assesses AI initiatives before any partnership or deployment. The option that plans the tutor on learners' own devices gets the process and the grade limit right but leaves the learners without devices behind, which is the equity the indicator asks for.

*Indicator 2.6.2: Manage emerging opportunities and challenges to encourage equality and equity in addressing the needs of learners, school personnel and other stakeholders.*

*Sources: DO 003, s. 2026, Secs. VI.1.1, VII.1.1, VII.1.6, VII.1.7, VIII.3.5 and VIII.4.5.*

### Question 2

*Indicator 3.5.2, Learning assessment. Career Stage 2, Situational, Moderate.*

You are the school head of a large national high school. A Grade 10 English teacher gave three learners' essays a score of zero after an online AI detector rated them "92% AI-generated." The learners say they wrote the essays themselves, and one parent has filed a complaint. The class had not been told whether AI could be used for the essay. What is the most appropriate action?

- **A.** Uphold the zero scores, since the detector's rating is high, and ask the parents to sign a pledge that their children will not use AI tools again.
- **B.** Coach the teacher to hear each learner's side and check the work through an in-class writing task, then set and explain the AI rules for future essays.
- **C.** Ask the teacher to tell the class that AI may not be used in future essays, and leave the three zero scores to the teacher's own judgment.
- **D.** Coach the teacher to give the essays back for revision with a reminder about AI use, and record the scores that the learners earn on the revised essays.

**Answer and explanation**

**Best answer: B.** Points: A = 0, B = 3, C = 1, D = 2.

- **A (0 points, Prestructural).** It makes a detector's rating the sole basis for a finding of dishonesty, which the guidelines forbid, and it punishes learners who were never told the rules.
- **B (3 points, Relational).** It lets learners respond, verifies authorship through classroom-based work as the guidelines direct, and fixes the root cause by setting clear rules for AI use.
- **C (1 point, Unistructural).** It sets a rule for future essays, one element the class needs, but it leaves zero scores standing that rest on nothing but a detector's rating.
- **D (2 points, Multistructural).** It drops the unfair zeros and reminds learners about AI, but it never verifies authorship and leaves the class without clear rules for AI use.

**Why.** This aligns with Career Stage 2 expectations because the school head gives the teacher technical assistance on fair assessment practice. The guidelines state that "AI detectors shall not be used as sole evidence of academic dishonesty" and call for in-person performance tasks and classroom-based activities to verify authenticity; teachers must also tell learners when, how and to what extent AI may be used. The option that returns the essays for revision drops the unfair scores but never checks whose work they are.

*Indicator 3.5.2: Provide technical assistance to teachers in using learning assessment tools, strategies and results consistent with curriculum requirements to ensure accountability in achieving higher learning outcomes.*

*Sources: DO 003, s. 2026, Secs. VI.1.2(ii), VI.1.2(v), VI.1.7(ii) and VIII.5.2.*

### Question 3

*Indicator 2.1.1, Records management. Career Stage 1, Conceptual, Easy.*

A school's registrar wants to paste the School Form with learners' names and grades into a free AI chatbot to produce a summary. Which statement best explains the problem with this plan?

- **A.** The chatbot's summary may contain errors, so the registrar would still need to check every figure against the original form.
- **B.** Such platforms may keep or train on what is entered, so learners' personal data would leave the school's control and protection.
- **C.** School Forms may be prepared only by hand, so a summary produced with any software would not be accepted by the division office.
- **D.** The registrar would have to declare the chatbot's use, so the summary would take longer to finish than a summary written by hand.

**Answer and explanation**

**Best answer: B.** Points: A = 2, B = 3, C = 0, D = 1.

- **A (2 points, Multistructural).** It names a real duty, checking AI output against the source, and the real risk of errors, but it treats the plan as acceptable once checked; even an accurate summary would breach the rule on personal data.
- **B (3 points, Relational).** It links the prohibition to its reason: documents with personally identifiable information may not go into AI platforms, especially those that store or train on inputs, because the data would leave the protection the Data Privacy Act requires.
- **C (0 points, Prestructural).** It is a misconception. Nothing bars software from school records work, and the guidelines allow AI to help with reports and dashboards when no personal data goes in.
- **D (1 point, Unistructural).** Declaring AI used on data is a real duty, but the claim that it makes the work slower is a misconception, and disclosure does not make it acceptable to enter personal data.

**Why.** This aligns with Career Stage 1 expectations because an aspiring school head must understand how to manage school data with technology. The guidelines provide that documents containing personally identifiable information "shall not be uploaded, shared, or processed through AI platforms, especially those that collect, store, or use input data for model training", in line with the Data Privacy Act of 2012. The option about checking for errors names real duties but misses the reason for the prohibition.

*Indicator 2.1.1: Demonstrate skills in managing school data and information using technology, including ICT.*

*Sources: DO 003, s. 2026, Secs. VI.1.3, VI.3.3 and VII.1.1.*

Nothing in this guide reproduces or predicts examination content.

## Frequently asked questions

### Can DepEd teachers use ChatGPT or Gemini?

Yes, as a support tool. DO 003, s. 2026 bans uses, not brands: a teacher may use a chatbot to prepare lessons, draft materials and tests, and draft feedback, checking every output, declaring the use, and entering no personal information about learners or staff (Secs. VI.1.1, VI.1.2, VI.1.7(ii) and VII.1.1). The DepEd AI Registry in Annex E lists ChatGPT and Google Gemini as high risk and Microsoft Copilot as limited to high risk, all three as pilots.

### Can Grade 1 to Grade 3 learners use AI apps in school?

No. "Direct learner engagement with AI tools shall be restricted to Key Stages 2 to 4 (Grades 4 to 12)", and in Kindergarten to Grade 3, AI is "limited exclusively to teacher preparation and the development of differentiated instructional materials" (DO 003, s. 2026, Sec. VI.1.1). Teachers of those grades may still use AI to prepare their own materials.

### Can a teacher give a zero because an AI detector flagged an essay?

No. "AI detectors shall not be used as sole evidence of academic dishonesty", and teachers are to verify authenticity through "in-person performance tasks and classroom-based activities" (DO 003, s. 2026, Sec. VI.1.2(v)). Learners must also have the chance to respond to AI-generated findings about their work.

### Do learners have to declare that they used AI?

Yes, for AI-assisted work they submit. Teachers make sure such submissions name the AI tool, the help it gave and the learner's own understanding of the work, and undisclosed use "may constitute academic dishonesty" under school policy (Sec. VI.1.7(ii)). Non-disclosure is not by itself a punishable offense; repeated or intentional misrepresentation may be (Sec. VI.1.7).

### Does DO 003, s. 2026 apply to private schools?

Not directly. "Private schools may adopt these guidelines consistent with their objectives and existing laws and regulations" (Sec. II). The Data Privacy Act and the child protection laws the order cites apply to private schools on their own terms.

## Sources

**DepEd issuances**

- [DepEd Order No. 003, s. 2026](https://www.deped.gov.ph/2026/02/20/february-20-2026-do-003-s-2026-foundational-guidelines-on-artificial-intelligence-ai-in-basic-education/), Foundational Guidelines on Artificial Intelligence (AI) in Basic Education, February 20, 2026, items 1 to 6 and Secs. I to XII, with Annex A (AI use declarations), Annex B (risk-based classification and sample use cases), Annex C (Digital Maturity Assessment Tool), Annex D (Privacy Impact Assessment guide) and Annex E (DepEd AI Registry).
- [DepEd Order No. 015, s. 2026](https://www.deped.gov.ph/2026/06/04/june-4-2026-do-015-s-2026-revised-guidelines-on-classroom-assessment-grading-system-and-awards-and-recognition-for-the-k-to-12-basic-education-program/), Revised Guidelines on Classroom Assessment, Grading System, and Awards and Recognition for the K to 12 Basic Education Program, June 4, 2026, Sec. V.C, paras. 31 to 38, Tables 5 and 6.
- [DepEd Order No. 016, s. 2026](https://www.deped.gov.ph/2026/06/04/june-4-2026-do-016-s-2026-guidelines-on-lesson-planning-and-learning-design/), Guidelines on Lesson Planning and Learning Design, June 4, 2026, paras. 23 to 25 and Annex B.
- [DepEd Order No. 006, s. 2026](https://www.deped.gov.ph/2026/03/24/march-24-2026-do-006-s-2026-guidelines-on-ensuring-a-safe-and-motivating-learning-environment-esmle/), Guidelines on Ensuring a Safe and Motivating Learning Environment, March 24, 2026, Sec. VI.5.g.
- [DepEd Order No. 017, s. 2026](https://www.deped.gov.ph/2026/06/04/june-4-2026-do-017-s-2026-strengthened-senior-high-school-curriculum/), Strengthened Senior High School Curriculum, June 4, 2026, Annex A (list of electives).
- [DepEd Memorandum No. 021, s. 2026](https://www.deped.gov.ph/2026/03/19/march-19-2026-dm-021-s-2026-2026-national-schools-press-conference/), 2026 National Schools Press Conference, March 19, 2026, para. 11 and the photojournalism rules, item 4.
- [DepEd Order No. 024, s. 2020](https://www.deped.gov.ph/2020/09/07/september-7-2020-do-024-s-2020-national-adoption-and-implementation-of-the-philippine-professional-standards-for-school-heads/), National Adoption and Implementation of the Philippine Professional Standards for School Heads, September 7, 2020, for indicators 2.1.1, 2.6.2 and 3.5.2.

**Laws and National Privacy Commission rules**

- [Republic Act No. 10173](https://lawphil.net/statutes/repacts/ra2012/ra_10173_2012.html), Data Privacy Act of 2012.
- [NPC Circular No. 2023-06](https://privacy.gov.ph/wp-content/uploads/2024/03/NPC-Circular-Repeal-16-01-Signed.pdf), Security of Personal Data in the Government and the Private Sector, December 1, 2023, Sec. 5.
- [NPC Advisory No. 2024-04](https://privacy.gov.ph/wp-content/uploads/2025/02/Advisory-2024.12.19-Guidelines-on-Artificial-Intelligence-w-SGD.pdf), Guidelines on the Application of Republic Act No. 10173 or the Data Privacy Act of 2012 (DPA), Its Implementing Rules and Regulations, and the Issuances of the Commission to Artificial Intelligence Systems Processing Personal Data, December 19, 2024, Secs. 1 and 2.B.

**News, announcements and commentary**

- Sevillano, S. "AI in classrooms? Philippine education department issues 'responsible use' guidelines." Philippine Daily Inquirer, via [Asia News Network](https://asianews.network/ai-in-classrooms-philippine-education-department-issues-responsible-use-guidelines/), February 26, 2026.
- GMA News. "[DepEd backs new national AI hub, cites safeguards in school integration](https://www.gmanetwork.com/news/topstories/nation/978023/deped-backs-new-national-ai-hub-cites-safeguards-in-school-integration/story)." February 26, 2026.
- GMA News. "[DepEd, Google to provide over 1 million teachers, staff with AI skills training, certification opportunities](https://www.gmanetwork.com/news/topstories/nation/990219/deped-google-teacher-training-ai/story/)." June 4, 2026.
- Cariaso, B. "[DepEd preparing master plan for AI in schools](https://www.philstar.com/cebu-news/2026/08/05/2547129/deped-preparing-master-plan-ai-schools)." The Philippine Star, August 5, 2026.
- Microsoft. "[DepEd and Microsoft Partner to Advance Philippine Education in the AI Era, Equip 1M Teachers with Microsoft Copilot](https://news.microsoft.com/source/asia/2026/08/27/deped-and-microsoft-partner-to-advance-philippine-education-in-the-ai-era-equip-1m-teachers-with-microsoft-copilot/)." August 27, 2026.
- Javier, D. R. C. [Contextualizing the K–12 Foundational Guidelines on Artificial Intelligence for Philippine Classrooms](https://www.pnuresearchportal.org/wp-content/uploads/2026/06/Policy-Brief-10-S-6_2026.pdf). Policy Brief Series 10(6). Philippine Normal University Educational Policy Research and Development Office, June 2026.

*Written by Mark Anthony Llego for NASH Reviewer. Sources: DepEd Orders No. 003, 006, 015, 016 and 017, s. 2026; DepEd Memorandum No. 021, s. 2026; RA 10173; NPC Circular No. 2023-06 and NPC Advisory No. 2024-04; the Philippine Daily Inquirer, GMA News, The Philippine Star, Microsoft and the Philippine Normal University.*

---

© 2026 Mark Anthony Llego, NASH Reviewer. All rights reserved. You may quote up to 100 words from this article with the author's name, its title and a link. In research, cite it as: Llego, M. A. (2026, November 9). *DepEd AI Guidelines: What DO 003, s. 2026 Allows and Bans*. NASH Reviewer. https://nash.com.ph/blog/deped-ai-guidelines-do-003-s-2026. Any other reuse of the article, its charts or its data needs written permission: email support@nash.com.ph. Terms of Service: https://nash.com.ph/terms#website-content-and-data