Anthropic to share its restricted cybersecurity AI model with EU
What's the story
Anthropic, a leading artificial intelligence (AI) firm, is giving the European Union (EU) access to its most advanced AI model, Mythos. The move comes after months of requests from the bloc over cybersecurity concerns. The European Commission confirmed that it had "several productive meetings" with Anthropic on this matter. "We welcome the latest developments on potential future access," said Thomas Regnier, an EU tech sovereignty spokesperson.
Model launch
Mythos was launched in April
Mythos was first launched in April, but only to a select few companies as part of Anthropic's cybersecurity initiative, Project Glasswing. The model is effective at detecting security flaws and vulnerabilities in software. Its launch raised concerns over potential cybersecurity threats from malicious actors. "Let's not forget that Mythos is not one off, a new wave of powerful models are coming to the market," Regnier said.
Access negotiations
Official terms of the deal are still unclear
The official terms of the deal between Anthropic and the EU are still unclear. Reports indicate that Anthropic has told the Commission it needs government permission to share Mythos, prompting the EU to step up talks with the US administration. The move comes as governments, banks, and tech firms are alarmed by the risks posed by powerful AI models like Mythos.
Model risks
The Trump administration had also announced agreements with AI firms
The release of Mythos has sparked fears over its potential to expose thousands of unknown software vulnerabilities. This could be exploited by malicious actors, thereby increasing the risk of cybercrime. The Trump administration had also announced agreements with Google DeepMind, Microsoft, and Elon Musk's xAI in May. These deals would allow the government to evaluate AI models before their public release.