Anthropic (ANTH.PVT) is expanding access to its carefully guarded Mythos model to approximately 150 additional organizations, the artificial intelligence lab said Tuesday, bringing the count of organizations with access to the model up to roughly 200.
The new organizations, which were not named in the announcement, are spread throughout sectors that “weren’t well-represented in our initial cohort,” Anthropic said, including power, water, communications, healthcare, and hardware.
Many of the new partners, which are based in more than 15 countries, are vendors of codebases relied upon by governments and other major national organizations, the AI lab said.
Anthropic has said it is allowing key organizations access to the Mythos model so that they can use the model to harden their own cybersecurity defenses. The model’s release has sparked concerns about the potential hacking capabilities of frontier AI models. Mythos reportedly possesses advanced cybersecurity and hacking abilities that could pose serious threats to national security and other critical infrastructure.
“Within 6 to 12 months, we expect that many other AI companies will have Mythos-class models, and they could release them without safeguards that prevent misuse,” Anthropic said. “We want Project Glasswing to spur institutions toward operating norms that reflect this reality.”
Early partners included the US government, leading investment banks such as JPMorgan Chase (JPM) and Goldman Sachs (GS), and a slew of major tech companies, including Amazon (AMZN), Apple (AAPL), Nvidia (NVDA), Broadcom (AVGO), and CrowdStrike (CRWD).
Anthropic CEO Dario Amodei appears at an event organized by the AI company. (Andrej Sokolow/picture alliance via Getty Images) · picture alliance via Getty Images
The initial cohort of roughly 50 partners has already found more than 10,000 “high or critical severity” security flaws, Anthropic said. Each new partner organization must pass a security clearance before gaining access to the model.
“What each partner has in common is that a successful attack on their codebase could be catastrophic,” Anthropic wrote. “For most partners, we estimate that a major attack could affect more than 100 million people, with important ramifications for both global and national security. “
Anthropic filed confidential paperwork with the Securities and Exchange Commission on Monday for a public offering, paving the way for the company to join SpaceX (SPAX.PVT) and fellow AI lab OpenAI (OPAI.PVT) in what are expected to be the largest and most closely watched IPOs of the year.
The AI lab announced last week that it had raised $65 billion in a Series-H funding round, bringing the company’s post-money valuation to $965 billion. Anthropic’s run-rate revenue recently crossed $47 billion, the company said.