Anthropic to offer EU access to its advanced Mythos model
This photo illustration shows the Antropik logo on a smartphone with the Claude mythos logo in the background.
Stick Pictures | Light Rocket | Getty Images
anthropic It is sharing access to its cutting-edge artificial intelligence model Mythos with the EU after the region sought access for months due to cybersecurity concerns.
The European Commission confirmed to CNBC on Monday that it had “several productive meetings” with the American artificial intelligence firm.
“We welcome the latest developments on potential future access,” EU technology sovereignty spokesman Thomas Regnier said in emailed comments, adding that the bloc aims to get a clearer idea of the potential risks posed by technology.
Anthropic first rolled out Mythos to a limited number of companies in April as part of Anthropic’s cybersecurity initiative, Project Glasswing, and the model has excelled at identifying security flaws and weaknesses in the software. The launch sparked a wave of concern about cybersecurity threats from bad actors.
“Let’s not forget that Mythos is not a one-off, a new wave of powerful models is coming to the market,” Regnier said. “This is a common challenge, and we are intensifying our discussions with like-minded partners, including the United States.”
The official terms and conditions of the deal are not yet clear. Anthropic did not immediately respond to CNBC’s request for comment.
The EU last week stepped up talks with the US administration to gain access after Anthropic told the Commission it needed government permission, people familiar with the matter previously told CNBC.
The U.S. government was opposed to sharing the powerful model with governments outside the U.S. because it aimed to remain the dominant AI leader, the people said at the time.
The EU was granted access to OpenAI’s GPT-5.5-Cyber model in May, but the Commission said at the time that it had held “four or five” meetings with Anthropic and that the agreements were “at a different stage”.
Governments, banks and tech firms were alarmed by the risks posed by Mythos when it was first launched in April; because this technology has the capacity to reveal thousands of previously unknown software vulnerabilities and can be misused in the hands of bad actors and accelerate cybercrime.
The Trump administration also announced agreements in May. Google deepmind, Microsoft and Elon Musk’s xAI, which will allow the government to evaluate AI models before they are released to the public.
