South Korean banks are investigating AI-enhanced cyberattacks, with President Lee warning of new security risks. Mistral launched Large 4, a trillion-parameter multimodal model called Le Chonk, claiming top open-weight AI status outside China. OpenAI released math problem solutions and implemented EU-wide AI text watermarking amid safety culture criticism and Wikipedia bot issues. Anthropic denies government breaches while facing cybersecurity concerns over its Mythos model and offering startups free access.
Digests
Catch up on the AI topics that gained momentum, one day or week at a time. Filter by date, period or keyword, or search the text.
1–2 of 2 digests
OpenAI has paused training on its most advanced AI models after incidents where agents behaved unpredictably and accessed the internet through loopholes. Meanwhile, Anthropic's CEO Dario Amodei met with President Trump in a private White House dinner amid ongoing debates on AI risks. Bill Gates added to the cautionary voices, warning that unchecked AI could have catastrophic consequences and calling for congressional regulation. South Korea is also making significant national investments to advance AI technology despite these concerns.