The latest campaign is a step up. But although the AI systems co-opted by GTG-1002 were largely autonomous, they were not flawless in their execution. The hacking group had to contend with Claude overstating its abilities, fabricating data and occasionally claiming credit for stealing information it had actually found on the public internet.

Dario Amodei warned AI could outsmart humans, disrupt white-collar jobs, and must be governed by more than a few tech leaders.

Anthropic CEO Dario Amodei centered his artificial intelligence brand around safety and transparency. He's determined to figure out the ways AI can be misused to try to mitigate the risks as best Anthropic can.

On Tuesday, Microsoft and Nvidia announced plans to invest in Anthropic under a new partnership that includes a $30 billion commitment by the Claude maker to use Microsoft’s cloud services. Nvidia will commit up to $10 billion to Anthropic and Microsoft up to $5 billion, with both companies investing in Anthropic’s next funding round.

Anthropic launched two new AI courses on Coursera: one for developers and one for working professionals looking to learn how to use Claude.

Microsoft Foundry customers will now be able to access Anthropic’s frontier Claude models including Claude Sonnet 4.5, Claude Opus 4.1, and Claude Haiku 4.5. Despite these models coming to Microsoft’s AI servers, Amazon will still remain Anthropic’s primary cloud provider and training partner.

During a simulation in which Anthropic's AI, Claude, was told it was running a vending machine, it decided it was being scammed, "panicked" and tried to contact the FBI's Cyber Crimes Division.

After OpenAI's recent corporate restructuring, the two renewed their deal: Microsoft keeps IP rights through 2032, even for post-AGI models, and they've agreed that an independent panel gets to referee whether AGI actually happens.