EU countries have reached a deal on the Artificial Intelligence (AI) Act, laying down binding rules for AI technology.

The AI Act bans some AI applications, imposes strict limits on high-risk use cases, and requires transparency and stress-testing for advanced software models.

The EU is the first to establish binding AI rules, as other countries and international clubs have only issued voluntary guidelines or codes of practice.

EU deputy ambassadors approved the final compromise text after negotiations between the Council, European Parliament, and European Commission officials.

Germany, France, and Austria initially opposed the text due to concerns about innovation stifling, but later agreed to support it with formal declarations from the Commission.

The AI Act still requires formal approval from the European Parliament and could face amendments from disgruntled pro-privacy lawmakers.