Anthropic blocked investigations that could have been related to biological weapons. According to a report titled "Detecting and Combating AI Misuse", published on Thursday, the company identified at least four suspected cases.
The cases were linked to state organizations. Anthropic decided not to disclose them publicly, considering the possibility that they could just be legitimate academic studies.
The report highlights the challenges that AI companies face when trying to distinguish between genuine academic research and potentially dangerous activities carried out by states.
This situation raises questions about the criteria used by AI companies to decide when and how they should report suspicions of misuse of their technologies.
