EU's cybersecurity agency granted access to Mythos 5 AI model, Commission says
The agency will use the restricted model to stress-test cyber defenses after Anthropic limited access to about 200 approved partners.
- On Thursday, the European Union's cybersecurity agency ENISA gained access to Anthropic's Mythos 5 model, with European Commission spokesperson Thomas Regnier confirming the agency is now testing it alongside OpenAI's GPT-6 Astra.
- Anthropic had restricted Mythos to roughly 200 partners via Project Glasswing, collaborating with Washington due to the model's ability to exploit cyber vulnerabilities with unprecedented speed. US export controls subsequently barred access for non-American users.
- Brussels pressed for access amid concerns that Washington might trigger a "kill switch" on advanced US technologies. European Commission holds significant regulatory powers under the EU AI Act, making companies wary of withholding information.
- Mythos took five months from Anthropic's original announcement to reach ENISA, ending protracted negotiations. European Commission spokesperson Regnier noted ENISA also holds access to OpenAI's GPT-5.6-Cyber model for testing.
- Regulators now utilize Article 55 to examine general-purpose AI models directly rather than relying on corporate disclosures. EU's AI Office began enforcement with a 36-person team to monitor systemic risks.
35 Articles
35 Articles
In today's Business Update, we will cover Gorrissen Federspiel's spectacular compensation case, Anthropic's "dangerous" AI model Mythos, and large companies' use of "AI rankings."
ENISA gained access to Mythos 5 for testing after lengthy negotiations with Anthropic.
The European Cybersecurity Agency (ENISA) is already testing the model, with very restricted access and so far concentrated mainly in the United States and a handful of partners. The agreement comes late on the June announcement and does not include the latest version, Mythos 5.1.
EU cybersecurity agency granted access to Anthropic's Mythos 5 model for testing
The European Commission has confirmed that the European Union Agency for Cybersecurity has been granted official access to Anthropic's restricted-access Claude Mythos 5 model.The high-tier AI...
Coverage Details
Bias Distribution
- 50% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
