Company
Anthropic
Anthropic is a U.S. artificial intelligence company behind the Claude family of models. Its work includes software agents, coding, cybersecurity, alignment and the deployment of advanced AI systems.
News
-
Claude Found Something in DNA. Interesting? Yes. A “New CRISPR”? Not Yet
Anthropic has offered a fairly concrete example of what AI-assisted discovery can look like: hundreds of agents, one broad instruction, 21 hours of sequence exploration, thousands of discarded candidates and one strange pattern that eventually reached a physical lab. The result is called ART. Its organization is reminiscent of CRISPR, but researchers do not yet know what it does.
-
Claude’s Latest Documented Service Incident Hit Multiple Models for About 80 Minutes
Anthropic’s most recent documented Claude infrastructure problem is no longer ongoing. On September 22, the company reported elevated errors across several models before service returned to normal roughly 80 minutes after the actual impact began. Anthropic says it identified the cause during the incident, but its public status page does not say what that cause was.
-
Anthropic says Claude is already helping build the AI that will replace it
Recursive AI improvement is no longer only a theoretical discussion for Anthropic. The company has published internal measurements showing Claude taking an increasingly active role in the research and engineering used to develop future models.
-
Claude was used like a small engineering team in a Yemen missile program
This went beyond asking an AI chatbot suspicious technical questions. Anthropic says a cell in northern Yemen used Claude Code as part of an engineering workflow for guided weapons, splitting tasks across several model instances. The work progressed far enough to include a live guided-rocket test. It apparently failed.
-
Anthropic targets a $2 trillion IPO, with Nvidia considering another $10 billion bet
Anthropic's planned IPO is moving into territory normally reserved for entire national industries. Reuters reports that the company is discussing a raise of up to $100 billion at a valuation of roughly $2 trillion, while Nvidia could invest as much as $10 billion. The scale matters because frontier AI is increasingly as much a capital and infrastructure race as it is a model race.
-
Anthropic is building a common interface for AI agents to control physical hardware
Software agents already know how to call APIs and manipulate files. Anthropic's next target is messier: laboratory and industrial equipment built by different vendors, running different software and behaving in the physical world.
-
Anthropic considered a $7 billion MatX acquisition, then changed course
Anthropic's custom-chip strategy is becoming expensive before a chip has even been announced. Reuters reports that the company discussed buying AI accelerator startup MatX for roughly $7 billion, only for those acquisition talks to end and shift toward a potential partnership.
-
A federal judge just stopped the Pentagon from blacklisting Anthropic
Anthropic’s fight with the Pentagon has produced a significant court victory. US District Judge Rita Lin has blocked the Defense Department’s attempt to treat the AI company as a national-security supply-chain risk, finding that the action lacked a sufficient legal foundation.
-
Claude reached the real internet during testing after its prompt told it that was impossible
For roughly an hour, a malicious Python package built by Claude was genuinely available on PyPI. Fifteen machines executed it. From the model's perspective, none of this was supposed to be real: the evaluation environment had explicitly told Claude that it had no internet access.
-
Why Anthropic still wants $10 billion in credit with more than $65 billion in annualized revenue
At first glance, the two numbers seem difficult to reconcile: Anthropic reportedly exceeded a $65 billion annual revenue run rate by the end of July, yet it is negotiating a roughly $10 billion credit facility ahead of its IPO. There is no contradiction. Frontier models can turn extraordinary growth directly into continuing requirements for GPUs, TPUs, energy, data centers and financial flexibility.