(June 1): Anthropic PBC is set to give the European Union’s cybersecurity agency access to Mythos, its powerful artificial intelligence (AI) tool capable of finding and exploiting vulnerabilities in computer systems.
The generative AI company is going to let ENISA join Project Glasswing, an initiative to let key organisations test Mythos capabilities before it’s released more widely, people familiar with the matter said.
Anthropic communicated the decision to the EU’s executive branch, the European Commission, over the weekend, the people said, asking not to be identified because the discussions are private. Commission officials travelled to San Francisco on Thursday to ask Anthropic executives for access to the model, following weeks of fruitless engagement.
A spokesperson for the commission said discussions with Anthropic were progressing and declined to comment further. A representative for Anthropic declined to comment.
Access to Mythos has been limited since it was first previewed in April due to fears that it could fall into the hands of cybercriminals. Anthropic has said the new model is extraordinarily adept at finding network vulnerabilities and could pose a major cybersecurity risk to critical systems.
Authorities in the US and some American financial institutions including Goldman Sachs Group Inc were given early access. The UK’s AI Security Institute also was allowed to test the model shortly after it was introduced.
See also: Anthropic eclipses OpenAI with valuation of US$965 bil
Uploaded by Chng Shear Lane
