Anthropic Expands 'Project Glasswing' for Enhanced AI Security Collaboration

AI company Anthropic announced the expansion of 'Project Glasswing,' increasing access to its Claude Mythos AI model from 50 to 200 organizations to bolster security across healthcare, power, water, and communication networks.

Wed Jun 03 2026

'Claude Mythos' Access Expanded to 200 Organizations

Anthropic, an AI company, has announced the expansion of its 'Project Glasswing' initiative, broadening access to its AI model, Claude Mythos. The project aims to increase participation from approximately 50 to nearly 200 organizations, fostering enhanced security collaboration through AI technology. The new participants hail from over 15 countries and represent diverse sectors including healthcare, power, water, and communication networks. Anthropic emphasized that participating organizations must meet strict security requirements before gaining access to Claude Mythos.

'Mythos' Model Specialized in Finding Software Vulnerabilities

Anthropic first unveiled its Mythos AI model in April, describing it as its most advanced system to date for coding and cybersecurity tasks. Initially, access was restricted to a select group of organizations, including Amazon, Microsoft, Nvidia, and Apple, through a controlled program called Project Glasswing. Anthropic collaborated with the US government, existing Glasswing partners, and cybersecurity industry members on the expanded rollout. The company stated that its technology has already helped partners identify over 10,000 serious or critically severe software flaws.

Potential Risks of AI Technology and Safeguards

The preview release of Mythos in April also raised concerns about the potential risks of advanced AI. The system demonstrated an ability to discover and combine small software bugs into more serious security threats. In response, US Treasury Under Secretary Scott Pesent and then Federal Reserve Chair Jerome Powell held an urgent meeting with bank chiefs to discuss the risks posed by the technology. Financial institutions sought access to the system to fix potential weaknesses in their own software. Anthropic plans to make Mythos class models available to all customers in the coming weeks, with additional safeguards in place.

*Source: YouTube: WION (2026-06-03)*

Tags: #ai #cybersecurity #tech #anthropic #software