Govt strengthens framework to curb AI‑generated deepfakes; mandates faster takedown
New Delhi, Aug 6
The Centre has enacted or strengthened rules including IT Rules to curb AI‑generated deepfakes, and reduced timelines for removal of unlawful content, an official statement said on Thursday.
Intermediaries are now required to take down unlawful information upon intimation within three hours instead of 36, and grievance redressal deadlines have been cut from 24 hours to two hours for sensitive cases such as nudity or impersonation, MoS, Electronics and Information Technology said Jitin Prasada in Lok Sabha.
The government remains conscious of the threats posed by deepfakes powered by Artificial Intelligence (AI), including synthetic audio, video and text, the minister said.
It has strengthened the regulatory framework through amendments to the IT Rules, 2021 to address harms arising from synthetically generated information (SGI), including deepfakes and AI-generated content.
Intermediaries are required to ensure clear labelling and traceable metadata for permissible AI-generated content, so that users can easily identify synthetically generated material and prevent deception or misuse.
It further strengthens user accountability and platform due diligence, including mandatory user awareness regarding legal consequences of unlawful AI-generated content and stronger compliance obligations for social media intermediaries.
The guidelines explicitly cover child sexual exploitation material, non-consensual intimate imagery, impersonation and other harmful AI-generated content, requiring platforms to prevent such content and take prompt action when detected.
Intermediaries must deploy reasonable and appropriate technical measures, including automated tools or other suitable mechanisms, to not allow any user to create, generate, modify, alter, publish, transmit, share, or disseminate any synthetically generated information that violates any law, the minister noted.
The minister mentioned initiatives related to deepfake detection including Saakshya, multi-agent framework developed by IIT Jodhpur & IIT Madras for deep fake detection; AI Vishleshak for improving Audio-Visual forgery detection system and IIT Kharagpur's project on Real-Time Voice Deepfake Detection System.
Intermediaries are required to appoint Grievance Officers and resolve complaints within the prescribed timelines.
The government has set up an appellate forum at the Central level to challenge decisions of intermediaries. Users can appeal online at this portal if their complaints are not addressed by the intermediaries' Grievance Officers.
SAHYOG Portal enables automated, centralized removal notices to intermediaries. Government and its authorised agencies across India use it to request removal of unlawful content.
— IANS
Reader Comments
It's reassuring to see India taking proactive steps against synthetic media. The time reduction from 36 hours to 3 hours is significant. However, I wonder how practical it is for smaller platforms to comply with these stringent timelines. Will there be adequate support and guidance for startups and smaller intermediaries? Technology is evolving so fast, enforcement needs to be smart, not just fast.
The labelling requirement is a great move! 👏 If AI-generated content is clearly marked, it becomes much harder to deceive people. But I'm concerned about the implementation - how many users actually check for labels or metadata? Also, these tools like Saakshya and AI Vishleshak - are they being made accessible to the public or just for law enforcement? Would love more transparency on that.
While this is a step in the right direction, I can't help but think about freedom of expression. With these sweeping rules, is there a risk of over-censorship? The definition of "unlawful content" can be quite broad. I hope the government ensures that genuine satire, parody, or artistic AI content isn't taken down. The appellate forum is good, but it needs to be truly independent and efficient.
As a PhD researcher in cybersecurity, I appreciate the government acknowledging the gravity of the deepfake threat. The IIT collaborations for detection are excellent initiatives. However, the cat-and-mouse game between deepfake creators and detectors is relentless. The real solution lies in digital literacy - we need mass awareness campaigns so that citizens can question what they see online. Technology alone won't save us.
We welcome thoughtful discussions from our readers. Please keep comments respectful and on-topic.