Trust in the government weaves the unconnected threads of our society into a single fabric. If the knit becomes excessively tight, it creates visible gaps and can even break the very threads that make the social fabric. The herculean struggle of Hyderabad’s students, environmentalists and civil society to protect the 400-acre Kancha Gachibowli forest site reveals a mighty fall in public trust in the Telangana government. While the government is now answerable to the Supreme Court for hurried bulldozing of trees and glaring procedural irregularities, it has now cracked down on users of AI-generated “fake news” on social media.
Story continues below this ad
The story of Kancha Gachibowli is that of a unique social movement where creative expression of dissent took an unprecedented leap with AI-powered visual storytelling using the abundantly available real footage of deforestation and distressed wild animals. Such a depiction that pressed for the state’s accountability was made possible with open-access generative AI platforms and the legal grey area that such technology occupies on questions of copyrights, transparency and liability.
AI content made with the intention to create misinformation, commit fraud, disturb public order, or the like are certainly matter to be pursued under the Bharatiya Nyaya Sanhita, 2023 and the Information Technology Act, 2000. In riot-hit Murshidabad, more than 1,000 social media accounts have been blocked by the state police in West Bengal for using AI in alleged criminal conspiracy and altering original videos and voices. In fact, if the state’s urge to remove anti-social threads online is stronger, there is more legal homework to be done after the Bombay High Court struck down Rule 3(1)(b)(v) of the amended IT Rules for not defining the terms “fake”, “false” or “misleading”, and violating Article 14 and 19 of the Constitution.
But the more pertinent question to ask for when the state is indeed at fault, as witnessed in the Kancha Gachibowli movement, is what legal framework could be for ethical and non-malignant use of AI media — an online creative activity that allows everyone a chance to be R K Laxman and Hayao Miyazaki of the post-Covid era.
Story continues below this ad
India is yet to pass laws that directly address AI as the core subject of regulation. Elsewhere, compulsory watermarking and labelling of deepfakes that disclose AI manipulation, with certain exceptions, are adopted as robust provisions under the European Union-AI Act, 2024 and its Digital Services Act, 2022. In the US, the Executive Order on Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence, 2023 includes mechanisms on labelling generative AI media and content provenance that traces its lifecycle. China’s Deep Synthesis Regulation, 2023 stipulates that deep synthetic content with a tendency to mislead the public must explicitly be labelled as such in a reasonable position and area.
While the full extent of norms remains to be realised, it is an evident legal principle that AI platforms must take responsibility in revealing the source and context of generative output. India’s Ministry of Electronics and Information Technology (MeitY) issued some advisories in the recent past on labelling synthetic media, but these legally non-binding communications were later withdrawn in parts.
As we await the legislature’s move on overall AI regulation, the texture of our social fabric is under duress with the ill-considered enforcement of public order. In Shreya Singhal v Union of India (2015), the Supreme Court adjudged the test for clear and present danger to public order and held, from an earlier judgement, that freedom of expression cannot be suppressed unless it leads to endangerment of community interest.
But in ground zero Telangana, Senior IAS officer Smita Sabharwal was issued a notice by the police to appear as witness for posting a Ghibli-style AI illustration of tree-felling in Kancha Gachibowli. The same image was reportedly also shared by 2,000 other individuals on the X platform. As the state hardens its hold selectively, there are three splits in our social fabric that stare at us.
One, it is impossible for law enforcement to trace the origin of generative AI and the number of users who share it without adopting content provenance measures under a new parent AI law. Two, applying existing penal provisions on misinformation upon bonafide use of unregulated AI, especially that which upholds community interests, is outside the current mandate of the police, regardless of political pressure. And three, attempting to prosecute only some AI users without a basis of fact or clear and present danger to public order is a violation of judicial precedents and, as Sabharwal terms it, a compromise of the principles of natural justice and equality before the law.
What dangerously damages each distinct thread of our social fabric is the kind of generative AI that substantially violates Article 19(2) and the privacy and dignity of individuals guaranteed under Article 21. Unlike the AI-generated content highlighting the destruction of the Kancha Gachibowli forest by the state, the more empirical threat to public order is the unabated use of deep fakes that completely disregard the truth, defile the conscience of the community and destroy everyday lives systematically. The Delhi High Court is hearing three civil petitions that seek immediate regulation of deep fake technology, for which the MeitY is slated to submit a comprehensive report in July this year.
Rightfully so, laws and rules are not born overnight. In the meantime, therefore, the state must observe caution before proceeding to prosecute AI users en masse, who are largely unaware of the evolving legal subject, and spare a moment to ease tensions in the threads, sew up leftover loops and strengthen the social fabric while earning back public trust.
The writer, an advocate at the High Court of Telangana, is associated with the Save City Forests collective
