Anthropic Expands Access to Mythos AI Cybersecurity Model Across 15 Countries

Anthropic is expanding access to its advanced cybersecurity-focused AI model, Mythos, adding 150 new partners across more than 15 countries through its Project Glasswing initiative. The expansion significantly broadens the reach of a model that has drawn attention for its ability to identify software vulnerabilities and security weaknesses. Anthropic said the latest group of participants includes sectors that were underrepresented during the program’s initial rollout, including power, water, healthcare, communications, and hardware.

The company said organizations joining Project Glasswing will be required to meet security standards before receiving access to the model. “This expansion is the next step toward our long-term goals: for AI to make all software more secure, and for us to help the industry adjust to how AI could change many of the core assumptions of cybersecurity,” Anthropic said in a blog post.

The announcement follows a series of developments for the AI startup. A day earlier, Anthropic said it would begin providing access to Mythos to the European Union after discussions with EU officials about the model’s cybersecurity implications. The company also recently confidentially filed paperwork with the Securities and Exchange Commission, beginning the process for a potential public offering.

Anthropic first introduced Mythos to a limited group of 50 partners in April through Project Glasswing. The model quickly became a topic of discussion among cybersecurity experts because of its ability to uncover previously unknown software vulnerabilities at scale.

While security researchers have long possessed similar capabilities, some experts warned that tools such as Mythos could allow malicious actors to identify and exploit vulnerabilities more rapidly if appropriate safeguards were not in place. The concerns prompted discussions between government officials, financial institutions, and major technology companies about how advanced cybersecurity-focused AI systems should be deployed and evaluated.

Anthropic said the early results from Project Glasswing have already demonstrated the model’s impact. According to the company, participating organizations have identified more than 10,000 high-severity or critical security flaws since the program launched. The company also pointed to the broader stakes involved in software security, estimating that a major cyberattack could affect more than 100 million people.

Several major technology and cybersecurity firms were among the original participants in Project Glasswing, including Apple, Nvidia, Microsoft, CrowdStrike, and Palo Alto Networks. Anthropic did not disclose the identities of the newly added partners.

By extending access to additional industries and international organizations, Anthropic is positioning Project Glasswing as a larger collaborative effort focused on evaluating how advanced AI systems can be used to strengthen cybersecurity while addressing concerns about potential misuse. The latest expansion gives a wider group of organizations the opportunity to test and assess Mythos as governments and businesses continue examining the risks and benefits of increasingly capable AI security tools.

This analysis is based on reporting from CNBC.

Image courtesy of techzine.eu.

This article was generated with AI assistance and reviewed for accuracy and quality.