3 months ago
1
The communicative truthful far: The Ministry of Electronics and Information Technology (MeitY) earlier this week notified an amendment to the IT Rules, 2021, that would necessitate users and societal media platforms to statement AI-generated content, and tighten the takedown timelines for each contented — not conscionable AI-generated posts — from 24-36 hours to 2 to 3 hours. The rules travel into effect connected February 20.
Also Read | IT Ministry mandates statement for AI-generated content, reduces takedown timeline to 2–3 hours
What astir AI-generated content?
The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026 necessitate societal media platforms to “prominently” statement “synthetically generated” content, oregon AI-generated images and videos. The request was archetypal projected successful October past year, and was notified this week. Social media platforms with much than 5 cardinal users are required to get a “user declaration [for AI-generated content] and [conduct] method verification earlier publishing [AI-generated content]”.
MeitY said successful an explanatory papers that this request was introduced to antagonistic “deepfakes, misinformation, and different unlawful contented susceptible of misleading users, causing users harm, violating privacy, oregon threatening nationalist integrity,” and that it was important that users beryllium alert whether what they are viewing is inauthentic.
The October draught explanation of “Synthetically Generated Information” (SGI) was wider, encompassing immoderate audiovisual contented that was AI-modified oregon generated. The last rules carve retired immoderate exemptions: for instance, smartphone-clicked photos that are retouched automatically by the camera app volition not beryllium considered SGI, and peculiar effects successful films volition not beryllium considered thing that needs to beryllium labelled. The rules besides prohibit definite types of SGI: kid intersexual exploitation and maltreatment material, forged documents, accusation connected processing explosives, and deepfakes falsely representing a existent person.
Also Read | AI labelling rules nearing finalisation, says IT Secretary S. Krishnan
How tin AI-generated contented beryllium detected?
The authorities has asked ample platforms to “deploy tenable and due method measures to forestall unlawful SGI, and to guarantee labelling/provenance/identifier requirements for permissible SGI”. A elder authoritative astatine the IT Ministry argued that ample platforms person blase tools to observe SGI, and that this request simply requires them to execute detection that they are already doing. Additionally, immoderate AI firms and platforms person participated successful the Coalition for Content Provenance and Authenticity (C2PA), which offers method standards to invisibly statement AI-generated contented successful a mode that tin beryllium work by different platforms, successful lawsuit AI-based detection fails. The rules allude to this effort by requiring “provenance/identifier requirements,” but the authoritative said they don’t privation to endorse immoderate azygous effort, but would similar to formalise the aims of specified collaborations.
Also Read | India’s caller AI governance guidelines propulsion hands-off approach
How person clip limits changed?
The IT Rules alteration immoderate authorities authorities and constabulary officials to contented takedown notices nether Rule 3(1)(b), and users to nonstop successful grievances for “illegal” categories of contented enumerated successful the IT Rules. Those categories see misinformation, nudity, and threats to sovereignty. For some government- and court-issued takedown notices, the timelines person been reduced to 2-3 hours, portion for each different categories of idiosyncratic complaints (like defamation and misinformation), the effect timelines person been reduced from 2 weeks to 1 week. The timelines for responding to idiosyncratic reports (under Rule 3(2)(b)) connected “sensitive” contented has besides been slashed from 72 hours to 36 hours. The authorities reasoned that the erstwhile limits allowed a batch of harm to beryllium done adjacent wrong those timelines, necessitating a revisit of the clip platforms person to act.
EDITORIAL | Too fake to beryllium good: On AI-generated imagery, labelling
What different changes person been made?
Users volition present person to person a reminder of platforms’ presumption and conditions much often. “The amendments see revisions to Rule 3(1)(c) of the Intermediary Rules, expanding the frequence of idiosyncratic notifications from erstwhile each twelvemonth to astatine slightest erstwhile each 3 (three) months, and expanding the contented of specified notifications to clarify imaginable consequences of non-compliance and reporting obligations,” JSA Advocates and Solicitors said successful an analysis.
The rules besides necessitate platforms to specifically pass users that harmful deepfakes and different amerciable AI-generated contented could exposure them to ineligible action, including the disclosure of their individuality to instrumentality enforcement agencies and “immediate disabling of entree oregon removal of specified content, suspension oregon termination of idiosyncratic accounts”, JSA said successful its analysis.
