Its work will include deliberately trying to make AI systems misbehave to expose flaws. Read More Post navigation Anthropic considers releasing new AI model ahead of IPO, sources say US clears way for US$2.7 billion sale of air-defence development upgrades to Ukraine