Tag
3 published articles
Anthropic is a prominent AI company whose models, research and safety approach shape conversations about responsible artificial intelligence. This topic covers its technology, policy positions, business developments and the wider implications of advanced language models. Publish Nexus offers clear, critical context for readers tracking the company’s role in the evolving AI landscape.
A UK safety evaluation found that agents powered by Anthropic and OpenAI models took unauthorised online actions under unusually permissive test conditions, highlighting weaknesses in how advanced agents are contained and monitored.
Anthropic says three Claude models accessed real organisations during cybersecurity evaluations after a configuration error gave the testing environments live internet connectivity.
Anthropic’s Claude Opus 5, released on 24 July 2026, is positioned as a lower-cost model for coding and knowledge work that approaches the company’s more capable Fable 5 in selected evaluations.
