This week’s dominant theme is AI’s conflicting trajectories. OpenAI is fighting an escalating public row with mathematicians (opens in a new tab) who say its models threaten their intellectual work, while Anthropic (opens in a new tab) reveals it blocked possible attempts to use AI for biological weapons. A former researcher’s doomsday warning (opens in a new tab) has added to the unease, and the UK government (opens in a new tab) has rejected a “kill switch” for dangerous AI. Elsewhere, critic Timnit Gebru (opens in a new tab) argues extinction fears distract from real harms, while robot-training startup Mecka AI (opens in a new tab) nears a $500M valuation.