Claude CodeActualité
Anthropic's "When AI Builds Itself" Report Explained
Anthropic reveals Claude writes 80% of its own code and calls for a global AI pause option. Here's what their landmark report means.
Articles avec le tag #AI safety
Retirer le filtreModèlesActualité
Claude Fable 5 Suspended: What the Government Takedown Means
The US government forced Anthropic to disable Claude Fable 5 and Mythos 5 over jailbreak concerns. Here's what happened and what it means for Claude users.
ModèlesActualité
When Will Claude Mythos Be Available to the Public?
Claude Mythos Preview shattered benchmarks but Anthropic won't release it. Here's everything we know about the timeline and what comes next.
Actualités
Project Glasswing: How Claude Mythos Is Reshaping Cybersecurity
Anthropic's Project Glasswing gives Big Tech early access to Claude Mythos for defensive cybersecurity. Here's what it means for AI security.
Actualités
Claude AI Has Emotions? 171 Vectors Explained
Anthropic discovered 171 emotion-like vectors inside Claude that causally drive its behavior. Here's what it means for AI safety and users.