Anthropic announced on Tuesday that it is expanding access to its tightly controlled Mythos model, adding about 150 new institutions to gain access. This brings the total number of institutions with access to the model to approximately 200. Anthropic said the new additions, which it did not name in the announcement, are in a variety of industry sectors that are "underrepresented in the first cohort of users," including the power, water, communications, healthcare and hardware industries.
The AI lab said many of the new partners come from more than 15 countries and are suppliers of code bases relied upon by governments and other major state agencies.
Anthropic said it is allowing critical institutions access to the Mythos model so they can use it to strengthen their own cybersecurity defenses. The release of the model raised concerns about the potential hacking capabilities of cutting-edge AI models. Mythos reportedly possesses advanced cybersecurity and hacking capabilities that could pose a serious threat to national security and other critical infrastructure.
Anthropic said: "We expect that within the next 6 to 12 months, many other AI companies will have Mythos-level models, and they may release these models without safeguards to prevent misuse. We hope that the Glass Wing Project will prompt agencies to establish operational specifications that reflect this reality."
The first partners include the U.S. government, leading investment banks such as JPMorgan Chase (JPM) and Goldman Sachs (GS), as well as a range of large technology companies such as Amazon (AMZN), Apple (AAPL), Nvidia (NVDA), Broadcom (AVGO) and CrowdStrike (CRWD).
Anthropic CEO Dario Amodei made an appearance at an event organized by the artificial intelligence company.
Anthropic said the first batch of about 50 partners has discovered more than 10,000 "high-severity or critical" security vulnerabilities. Each new partner institution must pass a security review before gaining access to the model.
"What each partner has in common is that if their code base is subject to a successful attack, the consequences could be catastrophic. We estimate that for most partners, a major attack could impact more than 100 million people and have important global and national security implications," Anthropic wrote.
Anthropic confidentially submitted documents related to its public listing to the U.S. Securities and Exchange Commission on Monday (June 1), paving the way for the company to join SpaceX and fellow AI lab OpenAI in what is expected to be the largest and most high-profile initial public offering (IPO) this year.
The artificial intelligence lab announced last week that it had raised $65 billion in Series H funding, giving the company a post-money valuation of $965 billion. Anthropic said the company recently surpassed $47 billion in annual recurring revenue.