2026-10-10

India's upload prompt: declare AI video before a big platform shows it

Under India's IT Rules, large social platforms must ask uploaders to declare synthetic media and verify it. What to answer, and what to log per Sume job.

4 min readSume

All posts

When you upload AI-made video to a large social platform that serves India, new Rule 4(1A) obliges the platform to make you declare whether it is synthetically generated, to check that declaration with technical measures, and to label it prominently if it is. So the practical job for a video team is to answer that prompt correctly, per file, and to keep the record that shows why. This is general information from the rule text, not legal advice.

What Rule 4(1A) says

The text is in the Ministry of Electronics and Information Technology's consolidated IT Rules, updated as on 10.02.2026, where Rule 4(1A) was inserted by G.S.R. 120(E) of 10 February 2026. It applies to a significant social media intermediary, which the rules define by a number of registered users in India above a threshold notified by the central government. A platform below that threshold is not under this sub-rule, so the prompt you see will vary by platform.

Rule 4(1A) in the consolidated rules, read 2026-10-10 from meity.gov.in
Step	What the platform must do, per the rule text
Before display	Require the user to declare whether the information is synthetically generated
Verify	Deploy appropriate technical measures, including automated tools, to verify the declaration, having regard to the nature, format and source of the information
Label	Where the declaration or verification confirms it, show a clear and prominent label or notice that it is synthetically generated
Consequence	If the platform knowingly permitted or failed to act on such content in breach of the rules, it is deemed to have failed due diligence
Scope of duty	Reasonable and proportionate measures to verify declarations and to see that no synthetic content is published without a declaration or label

How to answer the prompt

The rule's definition decides what counts. Synthetically generated information is audio, visual or audio-visual material made or altered with a computer resource so that it appears real and depicts a person or event in a way likely to be seen as indistinguishable from a natural person or real-world event. The same definition excludes routine or good-faith editing, formatting, colour adjustment, noise reduction, transcription or compression that does not materially alter substance, and uses solely for accessibility, translation or discoverability without altering any material part. The earlier post on routine editing goes through those exclusions.

Applied to a Sume pipeline, the sorting below follows the docs' own descriptions of each tool. Whether a given output is covered is still a judgement for you and your adviser.

Sorting Sume outputs against the rule's definition; tool descriptions from docs.sume.com, definition from meity.gov.in, read 2026-10-10
Output	How Sume docs describe it	Likely answer to the prompt
Text-to-video or image-to-video clip that looks like real people or places	Model generation through the video API	Declare it as synthetic
Avatar or talking-video of a realistic person	Generated by an avatar model	Declare it as synthetic
Trim of your own real footage	Worker ffmpeg only, no provider inference	Usually the routine-editing exclusion; keep the source
Filter or dim of real footage	Worker ffmpeg only, no provider inference	Usually routine editing; confirm it does not change meaning
Burned captions on real footage	Caption render, optional authored text	Accessibility exclusion if it only adds captions

Log it so the answer is repeatable

The Video 1.0 route accepts a metadata object that Sume stores on the job and does not send to the provider. Video 1.0 is documented as retiring soon, and the docs for POST /v1/videos do not list a metadata field, so keep the same declaration in your own job table too. Use the field to write your declaration at submit time, so the answer to the platform prompt comes from a record and not from memory. Example for the Video 1.0 route:

curl -X POST https://api.sume.com/v1/video-1.0/generate \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: in-upload-decl-001" \
  -d '{
    "prompt": "A street market at dusk, handheld",
    "duration": 6,
    "metadata": { "synthetic_declaration": "yes", "market": "IN" }
  }'

Sume docs do not state that generated files carry an embedded AI marker, so the platform's own detection may or may not recognise a clip. If a platform's check disagrees with your declaration, the job ID and the stored field show what you meant to say. Mixed files, such as real footage with one generated insert, are safer to declare as synthetic, since the rule's test is about what the material depicts. Ask your adviser if you want to answer otherwise.

Checklist

Use this as the pre-upload gate for clips going to Indian audiences.

Declare every realistic generated clip as synthetic when the prompt appears
Keep the original and the platform upload date next to the Sume job ID
Do not rely on file metadata to carry your AI status through the platform
Re-read the rule text and take advice for ads and for regulated categories

Sources

MeitY: IT (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, updated as on 10.02.2026 (read 2026-10-10)