Home / Technology / EU cybersecurity agency to test Anthropic AI
EU cybersecurity agency to test Anthropic AI
1 Jun
Summary
- Anthropic will grant EU cybersecurity agency access to its AI tool Mythos.
- Mythos can identify and exploit computer system vulnerabilities.
- Access is part of Anthropic's Project Glasswing initiative.
Anthropic PBC has decided to grant the European Union's cybersecurity agency, ENISA, access to its advanced artificial intelligence tool, Mythos. This AI is renowned for its capability to detect and exploit vulnerabilities in computer systems. ENISA will participate in Project Glasswing, an initiative designed for key organizations to test Mythos's capabilities prior to its broader public release.
This decision was communicated to the European Commission over the weekend. Commission officials had previously traveled to San Francisco to request access to the model. Access to Mythos has been restricted since its preview in April due to concerns about potential misuse by cybercriminals, as Anthropic has highlighted its significant cybersecurity risks.
Authorities in the United States and the UK's AI Security Institute, along with some American financial institutions like Goldman Sachs, have also received early access to test the model.
