Verizon Joins Anthropic’s Project Glasswing for AI Defence

Anthropic’s Project Glasswing has a new member, aiming to contain the complex vulnerabilities sniffed out by frontier AI – Verizon. 

The telecom giant partnered with Anthropic in its global initiative to protect critical infrastructure in the age of advanced AI. 

This would enhance Verizon’s ability to detect and patch vulnerabilities discovered by AI, while maintaining its long-standing commitment to network security and resilience.

Anthropic had launched Project Glasswing, following the introduction of Anthropic’s Claude Mythos Preview model, the company’s most powerful model yet and incredibly potent at identifying intricate vulnerabilities in critical software and operating systems at significant speed. 

Major players including AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorgan Chase, the Linux Foundation, Microsoft, NVIDIA and Palo Alto Networks are part of this coalition.

“Our customers rely on the security of our network every day. As part of Project Glasswing, we are able to test and improve our cybersecurity efforts with new insights to maintain our network’s security,” says Dan Schulman, Verizon CEO.

Claude Mythos Preview and the rise of AI threats

Owing to the many security implications of mass release, Claude Mythos Preview is currently a limited-release model.

Anthropic says that the system has already identified thousands of zero-day vulnerabilities, including some that have gone unnoticed for decades across major operating systems and browsers.

“Glasswing is built around Claude Mythos Preview, our new limited-release frontier model, which has so far found thousands of zero-day vulnerabilities – including some that survived decades of human review – spanning every major operating system and browser,” notes Daniela Amodei, President at Anthropic, on LinkedIn.

“AI cyber capabilities at this level will proliferate over the coming months and not every actor who gets access to them will be focused on defence. That's the gap Glasswing is built to close.

Daniela adds that: “Cyber defence at this scale is a team effort. Frontier labs, software companies, security researchers, open-source maintainers and governments all working together is how defenders will stay ahead.”

Building a global defence network

Project Glasswing is designed as a collaborative defence initiative, bringing together frontier AI developers, cybersecurity experts, software companies and governments. 

The goal is to ensure that advanced AI capabilities are used to strengthen, not weaken, global digital infrastructure.

“Over the past several months, our information security team has been rigorously testing this critical new technology to determine its benefits to our network,” Dan says. 

“By using Mythos Preview, we are uniquely positioned to share cross-industry insights that will help secure the global internet fabric and support our mission to deliver a safe and reliable experience for every customer.”

Verizon’s involvement reflects its broader strategy of integrating emerging technologies into its security operations while maintaining strict safety and evaluation standards. 

Every stage of testing with Claude Mythos Preview is governed by rigorous internal protocols to ensure responsible deployment.

The initiative also supports cross-industry learning, enabling participating organisations to share findings and best practices. This collective approach is intended to help security teams stay ahead of rapidly evolving cyber threats and reduce systemic vulnerabilities across critical systems.

As AI becomes more deeply embedded in cybersecurity operations, organisations are under increasing pressure to balance innovation with safety. Verizon has emphasised that all use of emerging AI technologies within its network is subject to strict governance and oversight.

Project Glasswing reflects a broader industry shift towards responsible AI adoption in high-risk environments. By combining frontier AI models with human expertise, organisations aim to improve detection capabilities while maintaining trust and accountability.