According to a recent LinkedIn post from Polymarket, Anthropic has introduced a cybersecurity-focused AI model called Claude Mythos through an initiative termed Project Glasswing. The post cites claims that the model can autonomously discover and exploit previously unknown zero-day vulnerabilities across major operating systems and web browsers.
Claim 30% Off TipRanks
- Unlock hedge fund-level data and powerful investing tools for smarter, sharper decisions
- Discover top-performing stock ideas and upgrade to a portfolio of market leaders with Smart Investor Picks
The post indicates that roughly 40 organizations, including Apple, Google, Microsoft, Amazon, Nvidia, CrowdStrike, JPMorgan Chase, and the Linux Foundation, are receiving early access, supported by $100 million in usage credits from Anthropic. It describes Mythos as demonstrating unusually high success rates in generating working exploits and performing complex vulnerability chaining, raising concerns about dual-use risks.
According to the content shared, Anthropic has briefed U.S. cybersecurity authorities such as CISA and other federal agencies, with at least one cited government source suggesting policymakers may be unprepared for the implications. The post frames this as a pivotal moment for AI safety and cybersecurity, suggesting that Anthropic is attempting to deploy powerful offensive-style capabilities in a defensive context before similar tools emerge from competitors or open-source ecosystems.
For investors, the post highlights both potential upside and risk for firms in AI, cloud, and cybersecurity that are named as early participants, as well as for incumbent security vendors facing a step-change in automated vulnerability discovery. The reference to a Polymarket market assigning a 28% probability that Mythos will be publicly released by June 30, 2026, offers a market-based gauge of expectations around broader commercialization and associated revenue opportunities.
If the capabilities described prove accurate, they could strengthen the strategic relevance of AI leaders like Anthropic in critical infrastructure protection and high-end enterprise security, potentially influencing future funding, partnerships, and regulatory scrutiny. At the same time, the emphasis on the model being “too powerful to release publicly” underscores regulatory, ethical, and liability questions that could affect valuation multiples for companies developing or relying on similar high-risk AI systems.