Three of the world’s biggest digital markets are trying to answer the same question: how do you control a technology that changes every few months, without killing it?
They’ve come up with very different answers. If you build software, run a website or simply use AI tools, these differences affect you more than you might think.
India: start with what people can see
India hasn’t passed a single big “AI Act.” Instead it went after the most visible problem first: deepfakes and AI-generated content.
Amendments to the IT Rules were notified on 10 February 2026 and came into force on 20 February 2026. They mainly deal with what the law calls “synthetically generated information,” meaning AI-made audio, video, images and text (Vaish Associates via Mondaq).
In practice, platforms now have to:
- Label AI content clearly, collect a user declaration at upload, and embed identifiers or metadata so the content can be traced (Entrepreneur India).
- Act fast on takedowns. Response windows have shrunk from 24-36 hours to 2-3 hours (CCIA explainer).
This sits alongside the Digital Personal Data Protection Act, which governs how personal data is collected and used.
What supporters say: deepfake scams, fake celebrity endorsements and morphed images were a real and fast-growing harm, and quick takedowns protect victims.
What critics say: tech industry groups have raised concerns about pre-publication checks and very short compliance windows (CCIA explainer). Some worry this could push platforms to remove lawful content just to be safe.
The European Union: the big rulebook, now with more time
The EU has gone furthest, with the AI Act, a full risk-based law. Some AI uses are banned outright. “High-risk” systems, such as AI used in hiring, education, credit or critical infrastructure, face heavy obligations around testing, documentation and human oversight.
But 2026 brought a major change. The original deadline for high-risk rules was August 2026. Under the Digital Omnibus on AI, which got final Council approval on 29 June 2026, standalone high-risk systems now have until 2 December 2027 (SecurePrivacy). AI built into regulated products such as medical devices and machinery gets until 2 August 2028 (Verdantix). The amendment was published as Regulation (EU) 2026/1744 and entered into force on 27 July 2026 (Forkast).
The banned practices were not delayed.
What supporters of the delay say: companies couldn’t comply without finished technical standards and guidance, so a delay was practical rather than a retreat.
What critics say: some commentators argue the delay takes the teeth out of the law and leaves people exposed to risky systems for longer.
The United States: who gets to make the rules?
The US has no national AI law. The real fight is over who regulates: Washington or the 50 states.
In December 2025, President Trump signed Executive Order 14365, “Ensuring a National Policy Framework for Artificial Intelligence.” It created a Department of Justice AI Litigation Task Force, which started work on 10 January 2026 with a mandate to challenge state AI laws in court (Paul Hastings). The order explicitly leaves out areas like child safety, data centre infrastructure and state procurement.
Meanwhile, the states kept going. By mid-2026, 29 states had enacted 109 AI laws (Model Diplomat). Colorado’s law, the first broad US high-risk AI regime, took effect on 1 February 2026 (Model Diplomat).
Many legal experts point out that an executive order on its own doesn’t override state law. That generally needs an act of Congress (Ropes & Gray).
What supporters of a single national standard say: fifty different rulebooks are expensive to follow and could slow American AI companies against global rivals.
What supporters of state laws say: with no federal law in place, states are protecting residents from real harms such as discriminatory hiring tools or chatbots aimed at children.
Side by side
| India | European Union | United States | |
|---|---|---|---|
| Main approach | Targeted rules on AI content and platforms | One comprehensive, risk-based law | No national law; federal-state tug of war |
| Biggest 2026 move | Deepfake labelling and fast takedowns (Feb 2026) | High-risk deadlines pushed to Dec 2027 / Aug 2028 | Executive order and DOJ task force against state laws |
| Who carries the burden | Social media and content platforms | AI developers and deployers | Depends on the state |
| Pace | Fast, rule-by-rule | Slow and detailed | Fragmented |
What this means for you
If you run a website or app in India, and it lets users upload or generate images, video or audio, check whether the new labelling rules apply to you. Even if they don’t, labelling AI content is a good trust signal for readers.
If you sell software to Europe, the high-risk deadline has moved, but don’t treat that as a holiday. The documentation and testing work takes a long time to do properly.
If you have US customers, look at state laws, not just federal news. The rules in Colorado or California may matter more than what happens in Washington.
If you’re an ordinary user, the visible changes will be more “AI-generated” labels on content and faster removal of fake videos.
If you’re building AI features and want them designed with compliance in mind, MindStick’s AI services team can help.
The bigger picture
No one has found the perfect balance between safety and innovation. India is moving fast on specific harms. Europe is building a comprehensive system and giving itself more time. America is still arguing over who’s in charge. Each approach has serious defenders and serious critics, and all three will keep changing.
The one safe prediction is that “we didn’t know the rules” won’t work as an excuse for much longer.
Which approach do you think is right for India? Share your view on MindStick YourViews. For the bigger debate about AI’s effect on society, see our posts on AI and electricity demand and AI browsers.
Disclaimer: This article is for general information only and is not legal advice.
