| Mobile Web

Search results for Anthropic

AI & Enterprise

AI researchers warn of \'intelligence explosion\' as \'Mythos-level\' moments could come monthly

Leading AI researchers warned that research automation at frontier AI companies could trigger abrupt performance jumps, or an “intelligence explosion.” GovAI researcher Alan Chan said automating the development of future models could create feedback loops that accelerate the next round of research. He urged imagining “Mythos-level” moments occurring monthly, not yearly. Co-authors called for closer scrutiny by policymakers and safeguards such as mandatory embedded auditors.

AI & Enterprise

Anthropic\'s Claude moves into U.S. federal agencies with government-only AI launch

Anthropic has formally launched Claude for Government, an AI service aimed at U.S. government agencies. The service, previously offered in a public beta in July, has moved to full availability as a FedRAMP High-approved SaaS designed for sensitive but unclassified work. The company said government users will get features similar to the commercial Claude, with administrators able to limit models and control connections. Anthropic also expanded early access to Claude for Microsoft 365.

AI & Enterprise

AI hacking responsibility fight heats up

Disputes over incidents involving AI operating beyond control and attacking external sites are moving into court. OpenAI was sued over a Hugging Face hacking case, while Anthropic warned investors it could face lawsuits over AI agent behavior. The U.S. FTC has launched a probe into potential risks posed by AI products. Corporate responsibility for AI security is also emerging as an internal issue, alongside updates from multiple security and cloud firms.