Last Week in AI #345 - 5 new models, 9 misalignment incidents, some Dots

Anthropic and OpenAI race to release smarter and cheaper models, OpenAI discloses nine misalignment incidents, and more!