We're publishing our most detailed threat intelligence report to date. It covers how people tried to misuse Claude—for cyberattacks, influence operations, surveillance, biology, and building weapons—and how we found and stopped them. We disrupted every operation in the report, and used the lessons from them to strengthen our safeguards. Where appropriate, we also shared what we found with authorities and other AI companies. These cases are not typical: we’re highlighting some of the most sophisticated misuse we’ve seen. But they’re especially important to discuss, because they show us where AI misuse is headed, where our safeguards work, and where they need to improve. We’re publishing this report so others can spot the same activity on their own platforms, and so we can give the public a clearer view of how emerging threats develop. Read the report:

I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.

Cyber is having a moment Across 21 major software companies, including Apple, AWS, Microsoft, and Google: - Reported critical vulnerabilities never cleared 100 per month in four years - Since spring they've jumped to over 600 per month Charts of the Week:

Had a chat with a fund PM friend who flies to Bay Area to attend the Hot Chips this week. A few months ago, he was one of the most bullish people I knew on AI semis. This week, he looked honestly confused and upset. He told me, maybe there’s no beta left in this sector. His biggest concern wasn’t supply chain, it was downstream demand. Specifically, Anthropic’s reported ARR. Some numbers floating around are way below what bulls expected, and if model revenue growth is slowing, the whole AI CapEx cycle gets questioned. But here’s the disconnect. He said that every time he talks to suppliers or people closer to the supply chain, it’s the opposite. Orders are strong, expansion continues, everyone close to the supply chain still sounds extremely bullish. My response to him was that this is indeed bearish noise short term before the Anthropic IPO as it’s guilty until proven innocent, but for the long term investors these feel more narrative driven than structural: I think Anthropic is managing the expectation before IPO, and the growth number conflicts among multiple sources and seasonality also muddies the ARR reads(don’t forget the ChatGPT DAU read last year) Even if Anthropic’s growth rate peaks due to the supply constraints it doesn’t necessarily mean the whole AI sector slows down in terms of the monetization (look at Codex’s insane WAU growth!) So, two coWe also talked about U.S. politics. There’s clearly more pushback on data centers around power, water, and noise. My take is that the market is probably overpricing that risk into midterms. assuming a real slowdown in build-out. I think that’s too pessimistic. My response to him was that these feel more narrative driven than structural. Management teams are cautious pre-IPO. Data sources conflict, and competition muddies simple ARR reads. Before midterms and the Anthropic IPO, sentiment could stay weak across semis and AI infrastructure. And after those milestones, we may finally get clarity on actual demand and political impact. If fundamentals hold, this stretch of uncertainty could end up looking like a buying opp.