Black October, explainedAppsGuidesAI DictionarySmall FixesOn RecordAboutFollow the lab

AI regulation

Also called EU AI Act, AI law, AI policy

AI regulation is the laws and rules governments set for how AI can be built and used, from outright bans to labelling AI content.

Safety and societyEveryday

Products, laws and numbers here change fast. Checked October 2026.

In plain words

Governments are deciding what AI should and should not be allowed to do. Most are not writing one giant rulebook for all AI. Instead they ask how risky a use is: an AI that sorts spam is treated very differently from one that decides who gets a job or a loan.

The European Union went furthest, with the AI Act. It bans a few uses outright, sets strict checks for high-risk ones, and requires chatbots and AI-made media to be labelled. The United States has no single national AI law and relies on existing laws, state laws and presidential orders. India has chosen guidelines plus existing laws over a separate AI act, while adding specific rules, such as labels on AI-generated content.

These rules change fast, so every fact on this page carries a date. We are not lawyers, and this is not legal advice.

Everyday example

Rules for vehicles, not for wheels

India does not have one law for wheels. A bicycle needs almost no paperwork, an auto needs a permit, a school bus needs fitness checks and a trained driver, and some vehicles are banned from highways. Rules follow the risk of the use. Most AI rules work the same way: the same technology faces light or heavy rules depending on what it is used for.

Where the example breaks: Vehicle categories are settled and easy to spot. With AI, the same model can be a harmless homework helper in one app and a hiring tool in another, so deciding which rules apply is often the hard part.

Go deeper

How it actually works

As of October 2026, here is where the three approaches stand. Dates are when rules apply.

  1. EU: the AI Act

    In force since 1 August 2024. Bans on practices like social scoring and harmful manipulation apply from 2 February 2025. Rules for general-purpose models like the ones behind chatbots apply from 2 August 2025. Labelling rules for chatbots and deepfakes apply from 2 August 2026.

  2. EU: the 2026 delay

    A follow-up law, the Digital Omnibus on AI, entered into force on 27 July 2026. It moved high-risk duties (for AI in hiring, credit, education and more) from August 2026 to 2 December 2027, and for AI inside regulated products to 2 August 2028. It also added bans on AI that makes fake intimate images of real people without consent, or child sexual abuse material, from 2 December 2026.

  3. United States

    No comprehensive federal AI law. A December 2025 executive order set up a federal effort to challenge state AI laws, and in March 2026 the White House asked Congress to pass a national framework that would override burdensome state rules. As of October 2026 Congress has not done so, and state laws such as California's frontier AI transparency law and Colorado's AI act (rewritten in May 2026, starting 1 January 2027) remain law, though some face court challenges.

  4. India

    MeitY released the India AI Governance Guidelines on 5 November 2025. They favour "innovation over restraint" and rely on existing laws like the IT Act and the DPDP Act rather than a separate AI law. They propose an AI Governance Group and an AI Safety Institute. Amended IT Rules from 20 February 2026 require labels on AI-generated content and faster takedowns.

  5. Other rules that apply anyway

    Privacy, consumer protection, copyright and anti-discrimination laws already cover many AI harms. See AI and copyright and AI and privacy.

EU risk tiers
Unacceptable (banned), high risk (strict checks), transparency risk (must disclose), minimal risk (no new rules, such as spam filters and video games)
India's seven sutras
Trust; People First; Innovation over Restraint; Fairness and Equity; Accountability; Understandable by Design; Safety, Resilience and Sustainability
US executive order
EO 14365, "Ensuring a National Policy Framework for Artificial Intelligence", 11 December 2025
Checked
28 October 2026. Rules change often: check official sources before relying on any date

Should AI get its own strict law now?

One view

Yes, set clear rules early

Risks like discrimination in hiring, deepfakes and unsafe products are already here. Clear rules protect people and give companies certainty. This is broadly the EU's approach.

The other view

No, use existing laws and adapt

AI changes fast, and detailed rules may age badly or slow useful innovation. Existing laws plus targeted fixes can handle most harms. This is broadly the approach of India's guidelines and current US federal policy.

Most likely: Both camps agree high-risk uses need oversight. They differ on timing and on how much to write down in advance. The EU's own delay of its high-risk rules in 2026 shows how hard the balance is. This is an open political question.

Questions people ask

What is the EU AI Act?

The European Union's law on AI, in force since August 2024 and applying in stages until 2028. It sorts AI uses by risk: some are banned, high-risk ones need checks and human oversight, and chatbots and AI-made media must be disclosed. It applies to companies outside the EU that serve EU users.

Does India have an AI law?

Not a dedicated one, as of October 2026. India relies on the IT Act, the DPDP Act and other laws, guided by the India AI Governance Guidelines of November 2025. IT Rules amended in February 2026 require labelling of AI-generated content on platforms.

Is there a US federal AI law?

No comprehensive one, as of October 2026. Federal policy is set mainly through executive orders and agency actions, while states like California, Colorado and Texas have passed their own AI laws. Congress has debated overriding state laws but has not passed a bill.

Does content made with AI need a label?

It depends on where you post and what it shows. In India, large platforms must ask users to declare AI-generated content and label it. In the EU, deployers of deepfakes must disclose them from August 2026. We are not lawyers; check the rules where you are.

When does the EU AI Act fully apply?

In stages. Bans from February 2025, general-purpose model rules from August 2025, transparency rules from August 2026, most high-risk rules from December 2027 and rules for AI inside regulated products from August 2028, after the 2026 Digital Omnibus delay.

Further reading

Prepared by Mundhra Labs. Reviewed by Aditya Mundhra. Checked October 2026.