What Exactly Are the New Rules?
In early 2026, India amended its Information Technology (IT) Rules from 2021 to specifically address what it calls “synthetically generated information” (SGI). This term covers any audio, image, or video created or altered by AI to appear authentic. The
rules mandate that all permissible AI-generated content must be clearly and permanently labeled so users know it's not real. Platforms are also required to embed traceable metadata to identify the content's origin where technically feasible. The most dramatic change is the takedown timeline. If AI-generated content is deemed unlawful, platforms like Facebook and YouTube must remove it within three hours of receiving a valid order from a court or government agency—a drastic reduction from the previous 36-hour window.
Who Do These Rules Target?
The regulations are aimed squarely at “intermediaries,” a broad category that includes social media platforms, search engines, and internet service providers. The goal isn't to punish individual users for creating AI memes but to hold the major distribution platforms accountable for the content they host. Significant Social Media Intermediaries—those with over five million users in India—face even stricter obligations. They must require users to declare if they are uploading AI-generated content and use technical tools to verify these declarations. Failure to comply can result in the loss of “safe harbor” protections, which shield them from liability for what their users post.
Why the Crackdown, and Why Now?
The Indian government's actions were spurred by a series of high-profile deepfake incidents that went viral, causing public outrage and raising alarms about the potential for misuse. These incidents, often involving celebrities and politicians, highlighted how easily AI could be used for harassment, fraud, and misinformation. With a massive and rapidly growing internet user base, authorities grew concerned about the technology's potential to disrupt elections or incite public disorder. The new rules are a direct attempt to get ahead of the problem by forcing platforms to take a more proactive role in content moderation.
Labeling, Takedowns, and Pre-Screening
The rules establish a two-pronged approach: transparency for permissible content and swift removal for illegal content. For anything AI-generated that doesn't break the law, the key is clear labeling. For content that is illegal—such as non-consensual intimate images, child abuse material, or impersonations intended to deceive—the focus is on speed. The timeline for addressing user complaints about sensitive content like nudity or impersonation has been slashed to just two hours. Furthermore, intermediaries are now expected to deploy “reasonable and appropriate” technical measures, including automated systems, to proactively detect and block the creation and spread of certain types of unlawful synthetic content before it even gets reported.
What This Means for Global Tech Companies
For U.S.-based tech giants like Meta, Google, and X (formerly Twitter), India's rules represent a significant operational challenge and a potential blueprint for regulation elsewhere. Complying with the aggressive three-hour takedown window requires immense resources and sophisticated automated systems. The advisories also initially suggested that “unreliable” or “under-testing” AI models would need government permission before deployment, a move that caused alarm in the industry. While this was later softened to an advisory rather than a binding rule, it signaled the government's intent to closely monitor AI development. These companies must now navigate a complex and demanding regulatory landscape in one of their largest markets, balancing compliance with concerns about free expression and innovation.
