Anthropic is expanding access to its cutting-edge artificial intelligence (AI) model "Mythos" to 150 corporations and institutions. Mythos is an AI model with expert-level security vulnerability detection capabilities that Anthropic unveiled in April.
Anthropic said on the 2nd (local time) that it is expanding the participants in Project Glasswing, a cybersecurity collaboration program based on Mythos, to 150 institutions in about 15 countries. Previously, Anthropic allowed access to Mythos only to about 50 institutions, including U.S. Big Tech corporations, citing the high risk of the model being misused for hacking.
The countries newly granted access to Mythos include Korea, Japan, Canada, Australia, New Zealand, France, Germany, Italy, Switzerland, the Netherlands, Spain, Belgium and India, the Financial Times (FT) reported, citing sources familiar with the matter.
Domestic corporations newly joining Project Glasswing include Samsung Electronics, SK hynix and SK Telecom, which have invested in Anthropic. The Ministry of Science and ICT is said to be participating through the Korea Internet & Security Agency (KISA).
Okta, Euroclear and Intercontinental Exchange (ICE), the operator of the New York Stock Exchange (NYSE), are also newly joining.
Anthropic said, "The institutions newly participating this time include sectors that were relatively underrepresented among the initial participants, such as power, water, health care, telecommunications and hardware," adding, "Many are vendors that provide products and services relied upon by countless institutions and governments worldwide, so a large-scale attack could affect more than 100 million people and have implications for national security."
Anthropic said it expanded access to Mythos to help governments and corporations proactively respond to new security threats and establish strong safeguards through Project Glasswing.
The company said, "Within 6 to 12 months, other AI corporations will also have security models on par with Mythos," adding, "They may release those models without safeguards to prevent misuse." It added, "In such an environment, cyberattacks could occur more frequently and in more unpredictable forms, and security personnel must devise response measures accordingly."
According to Anthropic, the institutions currently participating in Project Glasswing are using Mythos to detect and patch vulnerabilities, and also for pre-release security reviews that implement measures to prevent vulnerabilities from arising in the first place.
The announcement came a day after Anthropic was valued at $965 billion (about 1,460 trillion won) and confidentially filed a draft registration statement with the U.S. Securities and Exchange Commission (SEC) for an initial public offering (IPO). In the industry, this is seen as a move by Anthropic to expand the supply of security AI tools based on "Mythos," after aggressively securing corporate clients with its coding tool "Claude Code," which is based on its AI model "Claude."
In fact, Anthropic said, "We are providing tools we have developed to help Project Glasswing participants find vulnerabilities faster, upon request from each company's security team." The tools provided include "Claude Security," a codebase scanning tool that Anthropic recently unveiled.