Dossier
cybersecurity bugs
Coverage of cybersecurity bugs in the Nexus archive.
- AI on the couch: Anthropic gives Claude 20 hours of psychiatry
Anthropic released a 244-page system card detailing its new Claude Mythos model, described as their most capable yet, which is restricted to select companies due to cybersecurity concerns. The document explores philosophical questions about AI consciousness and intrinsic welfare as models grow more powerful.