This week saw major steps in AI ethics and safety worldwide.

The OECD launched a new system for companies to report AI risks. Over 40 countries will use this to spot dangers early. Kenya became the first African nation to release strict AI rules, including safety audits for tech firms.

At the UN, leaders debated stopping AI from spreading fake news in wars. A new study warned that synthetic data from AI could trick scientists if not marked clearly.

Engineers learned from IBM's failed health AI project – they now know to test AI tools better before using them. Military experts shared plans to keep humans in control of AI weapons at a big Geneva meeting.

Researchers got $500,000 grants to study super-smart AI safety. Companies like Google helped make rules for checking AI safety during development.