Anthropic published its second public Risk Report on Friday, disclosing that it has raised its assessment of catastrophic misalignment risk from "very low" to "low" and revealing an internal model it has no plans to release to the public. The…
Anthropic published its second Risk Report on Friday, upgrading its catastrophic misalignment risk rating from "very low" to "low" citing increased uncertainty from recent cybersecurity evaluation incidents.unite+1
The report discloses an unreleased internal model called Model 2 that outperforms Mythos 5 on many tasks; Anthropic says it has no plans to release it publicly.beincrypto+1
A months-long gap left biological safety classifiers inactive on all contractor traffic, though Anthropic says it found no evidence of misuse after remediation.unite+1