The guidelines also ask intermediaries to obtain user declarations on whether content is SGI
| Photo Credit:
Kesavan A N 1612@Chennai
Information Technology (IT) Rules draft amendments may not withstand legality test, warn experts.
Rights groups and industry bodies said the enforcement of proactive labelling and filtering responsibility on platforms and other significant intermediaries under the Draft Synthetically Generated Information (SGI) draft Rules of the IT Amendment Rules extend beyond the obligations of the IT Act itself.
The current draft raises concerns of the government stretching the IT Act to confer new powers upon itself, curbing freedom of expression in the process.
Taking issue with the term ‘intermediary,’ SFLC.in said that the guidelines do not clarify whether generative AI models would fall under the ambit of the definition of intermediary as under the IT Act, 2000.
clear definition
As per the Act, an intermediary clarifies that they store, host, transmit or receive content on behalf of another person. The intermediary cannot be the originator of content.
However, platforms such as ChatGPT, Perplexity, DeepSeek clearly generate content on the basis of prompts entered by users, rather than simply hosting or transmitting such content. Such platforms would not fall under the definition of intermediaries under the IT Act, and not be subject to regulation under the IT Rules, 2021.
“The intent of the amendment is clearly to also include content that is generated by the above-mentioned platforms. If they are excluded from the purview of this amendment, the effects of the same would be considerably reduced,” said SFLC.in pointing out that the Rules also lack exceptions where creation or alteration of information is required for various editing, creative work, etc.
content verification
The guidelines also ask intermediaries to obtain user declarations on whether content is SGI.
According to e Internet and Mobile Association of India (IAMAI), this puts the onus on platforms violating safe harbour norms. Earlier, the Supreme Court had noted that it is unreasonable and impracticable to expect intermediaries that receive millions of items of content daily—and numerous complaints—to judge the legitimacy of content. Given the volume of content, it is impracticable (if not impossible) to expect industry members to locate and identify all instances of SGI.
Moreover, intermediaries face severe penalties on failing to verify content. In such a scenario, the Internet Freedom Foundation warned that platforms are likely to over-label or block content if there’s any doubt.
“If an image is slightly edited, they might slap “SGI” on it (even if it’s just color-corrected by an algorithm). Or worse, they might prevent upload if their tool flags something (even if the tool is wrong). This can lead to significant censorship, including of political speech (imagine an opposition party’s genuine video being flagged as deepfake erroneously and delayed in posting during election time – it could affect outcomes),” said IFF.
Platforms might even resort to hashing images and comparing them to known AI outputs or scanning content. Such a surveillance method violates privacy “and possibly the spirit of Section 79 (no obligation to monitor)” said the IFF. It also sets a precedent where the government might next demand scanning for other categories eroding encrypted communications and user privacy.
Published on November 14, 2025
