Anthropic to grant EU cybersecurity agency ENISA access to Mythos AI: Report
In May, OpenAI gave access to its latest models including GPT-5.5-Cyber to Deutsche Telekom, BBVA
Anthropic is set to give its latest Claude Mythos Preview access to EU’s cybersecurity agency ENISA, a powerful tool with unprecedented capabilities to find vulnerabilities in software and web browsers.
As reported by Bloomberg, US-based artificial intelligence company ENISA to join Project Glasswing, an initiative which allows the major tech companies such as Google, Microsoft, and key businesses to test Mythos for finding critical vulnerabilities and securing operating systems.
According to sources privy to the matter, Anthropic has informed the European Commission of its decision regarding the request for access to its AI model.
This development follows a trip by Commission officials to San Francisco last Thursday, intended to secure access after weeks of unproductive negotiations.
Representatives for both the European Commission and Anthropic declined to disclose specific details. However a Commission spokesperson noted that discussions remain ongoing.
Since the release of Mythos, the model’s accessibility has been limited to certain companies or players because of the growing fears of misuse.
Even Anthropic CEO Dario Amodei called this model extraordinary which can not only find network vulnerabilities but also pose serious cybersecurity threats to financial institutions and critical systems.
Besides the EU, the major firms and financial institutions have been given early access to Mythos. The UK’s AI Security was also given access to test the model.
Two weeks earlier, an Anthropic spokesperson said in a statement, "We fully support our partners sharing findings with each other and companies outside of Glasswing to triage vulnerabilities.”
“While there was never a specific Glasswing NDA, confidentiality protections were something partners asked for at the outset and were built into agreements partners signed.”
The announcement comes weeks after the OpenAI gave access to its latest models including GPT-5.5-Cyber to Deutsche Telekom, BBVA and others European companies to help strengthen their resilience to vulnerabilities in their systems.
-
Adobe x Nvidia RTX Spark: Photoshop, Premiere get 2x AI speed boost
-
Why Nvidia CEO Jensen Huang calls AI job loss fears ‘complete nonsense’
-
Apple smart glasses delayed to late 2027: Report
-
Here's why London is becoming the next battleground for AI giants
-
Nvidia unveils Nemotron 3 Ultra: America’s smartest open-weights AI model, 30% cheaper to run
-
Nvidia unveils new CPU ‘superchip’ for Windows laptops: A new era for personal computers?
-
CNN sues Perplexity AI over alleged scraping of 17,000 news stories
-
MSI announces world’s first Agentic AI gaming monitor at Computex 2026
