Anthropic has decided to expand access to Claude Mythos Preview to approximately 150 additional organizations with a presence in more than 15 countries, thus reinforcing the reach of "Project Glasswing," the collaborative initiative that the 'startup' launched at the beginning of April with 50 initial partners with priority access to its AI model, including US authorities and companies.
"After several weeks of close collaboration with our Project Glasswing partners, the security industry, open-source software maintainers, and the US Government, we are extending the collaboration to approximately 150 new organizations," Anthropic announced, emphasizing that each of them must meet security requirements before being able to use the tool.
As explained by the company founded and led by Dario Amodei, the organizations incorporated in this phase are based in more than 15 countries and, for the most part, manage critical infrastructures. This expansion incorporates sectors that were barely present in the original group, such as energy, water, healthcare, communications, and hardware.
Furthermore, a good portion of the new participants are key suppliers: companies or non-profit organizations responsible for codebases on which countless public and private entities worldwide depend, including various governments.
"What all our partners have in common is that a successful attack on their source code could be catastrophic," Anthropic stressed, estimating that, for most of them, a serious intrusion could impact more than 100 million people and entail relevant consequences for global and national security.
"This expansion represents the next step towards our long-term goals: for AI to improve the security of all software and for us to help the industry adapt to how AI could transform many of the fundamental principles of cybersecurity," the company added.
Since Project Glasswing was launched in early April, Anthropic and around 50 initial partners with access to the preliminary version of Claude Mythos have used the model to review codebases and locate vulnerabilities, identifying more than 10,000 high-severity or critical security flaws to date.
Anthropic's new statement comes one day after the AI firm confidentially filed a draft registration statement with the U.S. Securities and Exchange Commission (SEC) for a potential initial public offering (IPO) of its common stock.
"Anthropic, PBC has confidentially submitted a draft registration statement on Form S-1 to the SEC for a proposed initial public offering of our common stock," the company confirmed.
With this step, Anthropic reserves the possibility of debuting on the stock market once the SEC concludes its analysis, although the company has emphasized that the eventual IPO will depend on market conditions and other external factors.
For the moment, neither the number of shares to be offered for sale nor the initial price range has been determined. In this regard, the firm insisted that this statement "does not constitute an offer to sell securities, nor a solicitation of an offer to buy them."
Anthropic's move joins the stock market race of major tech startups. At the end of May, it was reported that OpenAI, the creator of ChatGPT, was advancing its plans to go public in the fall, while SpaceX also filed the prospectus in May to list on the Nasdaq in New York and Texas under the symbol "SPCX," a deal that could become the largest IPO in history.
Last week, Anthropic announced that it had raised $65 billion ($55.895 billion euros) in a funding round led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, which valued the company at $965 billion ($829.831 billion euros), above OpenAI's market valuation.
In February, the company had already managed to raise $30 billion ($25.798 billion euros) in another round that then valued it at $380 billion ($326.733 billion euros).