Anthropic is expanding Project Glasswing, its initiative aimed at identifying software vulnerabilities with artificial intelligence, to around 150 additional organizations across more than 15 countries, the company said Tuesday.
The announcement came a day after Anthropic disclosed that it had confidentially filed for an initial public offering following a funding round that valued the company at nearly $1 trillion.
At the center of Project Glasswing is Claude Mythos, which Anthropic describes as its most powerful model.
According to the company, the model can uncover thousands of zero-day vulnerabilities over a period of several weeks.
Program broadens after initial rollout
In early April, Anthropic granted 50 initial partners, including the U.S. government, access to Claude Mythos Preview to scan codebases for security weaknesses.
The latest expansion adds organizations from sectors that were underrepresented in the original group.
Those sectors include:
- Power
- Water
- Healthcare
- Communications
- Hardware
Anthropic said many of the new participants maintain software that governments and other organizations depend on.
“What each partner has in common is that a successful attack on their codebase could be catastrophic,” the company said.
“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.”
Countries involved
According to the Financial Times, citing a person familiar with the matter, organizations receiving access are located in countries allied with the United States.
They include:
- Australia
- Belgium
- Canada
- France
- Germany
- India
- Italy
- Japan
- Netherlands
- New Zealand
- South Korea
- Spain
- Sweden
- Switzerland
Reported participants
The Financial Times also reported that several organizations have been granted access to Claude Mythos.
They include:
- Okta
- Samsung
- SK Hynix
- SK Telecom
- NATO
- ENISA, the European Union’s cybersecurity agency
TechCrunch said it had contacted Anthropic for confirmation.
Competition in AI cybersecurity
Anthropic has said it expects competing AI companies to develop models with capabilities similar to Mythos Preview.
The company says it is moving quickly to establish safeguards through Project Glasswing.
Rival OpenAI has also entered the field.
Since the release of Mythos, OpenAI has introduced GPT-5.5-Cyber, a model focused on cybersecurity.
OpenAI has made the system available to a broad group of partners for testing.
TL;DR:
Anthropic is expanding Project Glasswing to about 150 organizations in more than 15 countries. The initiative uses the Claude Mythos model to detect software vulnerabilities in sectors such as healthcare, power and communications.
AI summary:
- Anthropic is expanding Project Glasswing to 150 organizations.
- Participants span more than 15 countries.
- Claude Mythos is used to identify software vulnerabilities.
- New sectors include power, water, healthcare and communications.
- OpenAI has launched a competing cybersecurity model, GPT-5.5-Cyber.
