AI digest: Anthropic's busy week
Anthropic makes waves with a $400M biotech acquisition, new PAC, and desktop control features whilst OpenAI shuffles executives.
Anthropic dominated the headlines this week with major moves across acquisitions, politics, and product features.
Anthropic buys biotech startup for $400M
Anthropic acquired Coefficient Bio for $400 million in a stock deal. The biotech startup was operating in stealth mode, suggesting Anthropic is betting big on AI applications in biological research. This feels like a serious play for the life sciences market rather than just talent acquisition.
Claude takes control of your desktop
The company also rolled out desktop control features for Claude Code and Cowork. Users can now let Claude directly operate their Mac or Windows computers to handle routine tasks. We’re curious how this plays out in practice, desktop automation has been promised for decades but rarely works smoothly.
Anthropic launches political action committee
Not content with just technical developments, Anthropic formed a new PAC ahead of the midterm elections. They’re positioning to back candidates who support their policy agenda. Smart timing given the regulatory scrutiny all AI companies are facing.
DeepMind’s self-improving algorithms
Meanwhile, Google DeepMind published research on AlphaEvolve, an LLM that rewrites its own game theory algorithms and outperformed human experts. The system improved multi-agent reinforcement learning algorithms through evolutionary coding. This could be significant for any application where agents need to compete or cooperate strategically.
OpenAI executive changes
In comparison, OpenAI had a quieter week with just executive reshuffling. COO Brad Lightcap moves to lead “special projects” whilst CMO Kate Rouch steps away for cancer recovery. The “special projects” role suggests something interesting brewing behind the scenes.