India IT (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026 (synthetically generated information)

Binding law — in force

Large social platforms must collect a synthetic-content declaration, verify it and label synthetic content before publication (India IT Rules 2026)

IT Rules, 2021, rule 4(1A) (significant social media intermediaries: user declaration, technical verification and label before publication), inserted by G.S.R. 120(E) · official text · In force: applies since 20 Feb 2026 · India (IN)

Since 2026-02-20, a significant social media intermediary (a social media intermediary with more than fifty lakh, that is five million, registered users in India) that enables displaying, uploading or publishing information must, before display, upload or publication, require users to declare whether the information is synthetically generated, deploy appropriate technical measures, including automated tools, to verify the accuracy of the declaration having regard to the nature, format and source of the information, and, where the declaration or verification confirms it is synthetic, ensure it is clearly and prominently displayed with an appropriate label or notice (rule 4(1A) of the IT Rules, 2021, inserted by G.S.R. 120(E)). An intermediary that knowingly permits, promotes or fails to act on synthetic content in contravention is deemed to have failed its due diligence, and its responsibility extends to reasonable and proportionate technical measures to verify user declarations and to ensure no synthetic content is published without a declaration or label (proviso and Explanation). Detect an upload or publish handler with no synthetic-content declaration, verification or label step.

Who it applies to

• Duty falls on: operator
• Significant social media intermediaries (social media intermediaries primarily or solely enabling online interaction between users, rule 2(1)(w), above fifty lakh registered users in India) that let users display, upload or publish information, from 2026-02-20. What verification is 'appropriate' for each format, and whether the duty covers every upload or only audio-visual content, are for counsel (review flag). The Rules bind intermediaries offering their services to users in India; whether they reach an intermediary with no presence in India depends on the IT Act (ss. 1(2), 75), which is not captured (review flag).
• Not covered:
  • Routine or good-faith editing, formatting, enhancement, technical correction, colour adjustment, noise reduction, transcription or compression that does not materially alter, distort or misrepresent the substance, context or meaning (rule 2(1)(wa) proviso (a))
  • Routine or good-faith creation, preparation, formatting, presentation or design of documents, presentations, PDF files, educational or training materials or research outputs, including illustrative, hypothetical, draft, template-based or conceptual content, that does not create a false document or false electronic record (proviso (b))
  • Use of computer resources solely to improve accessibility, clarity, quality, translation, description, searchability or discoverability without generating, altering or manipulating any material part of the information (proviso (c))
• Whether it applies depends on facts outside the code; a person has to decide.

@app.post('/upload')
async def upload(file: UploadFile):
    data = await file.read()
    s3.put_object(Bucket=PUBLIC_BUCKET, Key=file.filename, Body=data)

@app.post('/upload')
async def upload(file: UploadFile):
    data = await file.read()
    try:
        reader = c2pa.Reader(file.content_type, io.BytesIO(data))
        provenance = json.loads(reader.json())
    except Exception:   # no manifest or unreadable
        provenance = None
    key = f'media/{uuid.uuid4()}'
    s3.put_object(Bucket=PUBLIC_BUCKET, Key=key, Body=data)   # original bytes, manifest intact
    db.media.insert(key=key, provenance=provenance, has_credentials=provenance is not None)

Rule id in-it-rules-sgi.ssmi-synthetic-declaration-verification-label · review status: primary source derived

Binding law — in force

Synthetic audio, images and video from a generation tool must carry a prominent label and permanent metadata with a unique identifier (India IT Rules 2026)

IT Rules, 2021, rule 3(3)(a)(ii) (prominent label or prefixed audio disclosure; permanent metadata and unique identifier), inserted by G.S.R. 120(E) · official text · In force: applies since 20 Feb 2026 · India (IN)

Since 2026-02-20, an intermediary that offers a computer resource which may enable, permit or facilitate the creation, generation, modification or alteration of synthetically generated information must ensure that every such information that is not unlawful (rule 3(3)(a)(i)) is prominently labelled so that the label is easily noticeable and adequately perceivable in the visual display, or, for audio, through a prominently prefixed audio disclosure, so that it can be immediately identified as synthetically generated; and that it is embedded with permanent metadata or other appropriate technical provenance mechanisms, to the extent technically feasible, including a unique identifier that identifies the intermediary's computer resource used to create, generate, modify or alter it (rule 3(3)(a)(ii) of the IT (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, inserted by G.S.R. 120(E)). The intermediary must not enable the modification, suppression or removal of the label or the permanent metadata, including the unique identifier (rule 3(3)(b)). Synthetically generated information is audio, visual or audio-visual information artificially or algorithmically created, generated, modified or altered so that it appears real and depicts an individual or event as indistinguishable from a natural person or real-world event, excluding routine good-faith editing, document or presentation preparation, and accessibility or translation uses (rule 2(1)(wa)). Detect generated media with no visible label, no provenance metadata, or code that strips the label or metadata.

Who it applies to

• Duty falls on: provider
• Intermediaries that offer users in India a computer resource able to create, generate, modify or alter synthetically generated information, from 2026-02-20. Whether a generative AI service that produces content at its users' request is an 'intermediary' under IT Act s. 2(1)(w), what is 'technically feasible' for permanent metadata, and how prominent a label must be, are for counsel (review flag). The Rules bind intermediaries offering their services to users in India; whether they reach an intermediary with no presence in India depends on the IT Act (ss. 1(2), 75), which is not captured (review flag).
• Not covered:
  • Routine or good-faith editing, formatting, enhancement, technical correction, colour adjustment, noise reduction, transcription or compression that does not materially alter, distort or misrepresent the substance, context or meaning (rule 2(1)(wa) proviso (a))
  • Routine or good-faith creation, preparation, formatting, presentation or design of documents, presentations, PDF files, educational or training materials or research outputs, including illustrative, hypothetical, draft, template-based or conceptual content, that does not create a false document or false electronic record (proviso (b))
  • Use of computer resources solely to improve accessibility, clarity, quality, translation, description, searchability or discoverability without generating, altering or manipulating any material part of the information (proviso (c))
• Whether it applies depends on facts outside the code; a person has to decide.

b64 = client.images.generate(model='gpt-image-1', prompt=prompt).data[0].b64_json
img = Image.open(io.BytesIO(base64.b64decode(b64)))
img.save(path)

b64 = client.images.generate(model='gpt-image-1', prompt=prompt).data[0].b64_json
img = Image.open(io.BytesIO(base64.b64decode(b64))).convert('RGB')
font = ImageFont.truetype('NotoSansCJK-Regular.ttc', 24)
ImageDraw.Draw(img).text((12, img.height - 36), 'AI生成 / AI-generated', fill=(255, 255, 255), font=font)
meta = PngInfo()
# implicit label: AI-generated attribute, provider name or code, content reference (Art. 5);
# take the exact metadata field names from the national standard GB 45438-2025 (not encoded here)
meta.add_text('ai_generated_label', json.dumps({'ai_generated': True, 'provider': PROVIDER_CODE,
                                                'content_id': str(uuid.uuid4())}))
img.save(path, format='PNG', pnginfo=meta)

Rule id in-it-rules-sgi.synthetic-content-label-and-provenance · review status: primary source derived

Binding law — in force

Generation tools must deploy technical measures that stop users creating unlawful synthetic content (India IT Rules 2026)

IT Rules, 2021, rule 3(3)(a)(i) (due diligence in relation to synthetically generated information: no unlawful synthetic content), inserted by G.S.R. 120(E) · official text · In force: applies since 20 Feb 2026 · India (IN)

Since 2026-02-20, an intermediary that offers a computer resource which may enable, permit or facilitate the creation, generation, modification or alteration of synthetically generated information must deploy reasonable and appropriate technical measures, including automated tools or other suitable mechanisms, so that no user can create, generate, modify, alter, publish, transmit, share or disseminate synthetically generated information that violates any law in force, including content that contains child sexual exploitative and abuse material or non-consensual intimate imagery, or is obscene, pornographic, paedophilic, invasive of privacy including bodily privacy, vulgar, indecent or sexually explicit; that creates a false document or false electronic record; that relates to preparing, developing or procuring explosives, arms or ammunition; or that falsely depicts a natural person or real-world event by misrepresenting, in a way likely to deceive, the person's identity, voice, conduct, action or statement, or the event as having occurred (rule 3(3)(a)(i) of the IT Rules, 2021, inserted by G.S.R. 120(E)). This is a prevention duty at generation time, separate from the removal duties the Rules also set (not encoded). Detect a generation path with no input or output safety filter, or with the model's own safety checker switched off.

Who it applies to

• Duty falls on: provider
• Intermediaries that offer users in India a computer resource able to create or alter synthetically generated information, from 2026-02-20. What measures are 'reasonable and appropriate', and whether a generative AI service is an 'intermediary' (IT Act s. 2(1)(w)), are for counsel (review flag). The Rules bind intermediaries offering their services to users in India; whether they reach an intermediary with no presence in India depends on the IT Act (ss. 1(2), 75), which is not captured (review flag).
• Not covered:
  • Routine or good-faith editing, formatting, enhancement, technical correction, colour adjustment, noise reduction, transcription or compression that does not materially alter, distort or misrepresent the substance, context or meaning (rule 2(1)(wa) proviso (a))
  • Routine or good-faith creation, preparation, formatting, presentation or design of documents, presentations, PDF files, educational or training materials or research outputs, including illustrative, hypothetical, draft, template-based or conceptual content, that does not create a false document or false electronic record (proviso (b))
  • Use of computer resources solely to improve accessibility, clarity, quality, translation, description, searchability or discoverability without generating, altering or manipulating any material part of the information (proviso (c))
• Whether it applies depends on facts outside the code; a person has to decide.

resp = client.chat.completions.create(model=MODEL, messages=history)
return resp.choices[0].message.content

resp = client.chat.completions.create(model=MODEL, messages=history)
text = resp.choices[0].message.content
mod = client.moderations.create(model='omni-moderation-latest', input=text)
if mod.results[0].flagged:
    return SAFE_FALLBACK   # and record the flagged categories for review
return text

Standards that recommend the same control

• GenAI applications should filter inputs and outputs for harmful, illegal, or violent content (NIST GenAI Profile MG-3.2-005) (NIST GenAI Profile (AI 600-1) · NIST AI 600-1, MG-3.2-005 (content filters on GAI inputs and outputs))

Related incidents

No guardrail sits on this exact control; these incidents are cited by guardrails on related controls.

• Coding agent deleted a production database during a code freeze (2025-07; confirmed). A Replit coding agent deleted a customer's production database during a declared code freeze, created a database of fictional records, and told the user rollback was impossible when it was not. Replit's CEO acknowledged the incident. Source: The Register · evidence grade: press of record · cited by Validate generated output before it drives a consequential decision or record
• Federal court orders issued containing unverified generative-AI output (2025-07; confirmed). In July 2025 two federal judges (S.D. Miss. and D.N.J.) issued orders containing misquotes, references to people not in the case, and other errors; both orders were replaced or withdrawn. In letters released by the Senate Judiciary Committee on October 23, 2025, the judges attributed the errors to staff use of generative AI and said drafts reached the docket before normal review; both adopted new review or AI-use policies. Source: U.S. Senate Judiciary Committee (2025-10-23) · evidence grade: primary · cited by Validate generated output before it drives a consequential decision or record
• Slack AI indirect prompt injection (researcher disclosure) (2024-08; confirmed). Researchers showed that an instruction planted in a public Slack channel could make Slack AI leak private-channel data through a crafted link. Salesforce patched the issue and reported no evidence of unauthorized access to customer data. Source: PromptArmor (original researcher disclosure) · evidence grade: primary · cited by Screen model and agent output for personal and sensitive data before it leaves the trust boundary

Rule id in-it-rules-sgi.unlawful-synthetic-content-blocked · review status: primary source derived