Verizon Joins Anthropic Project Glasswing Collaboration

Verizon announced on May 15, 2026 that it has joined Anthropic's Project Glasswing, per a Verizon news release published the same day. The release describes Project Glasswing as an initiative that grants controlled access to Anthropic's most advanced model, Claude Mythos Preview, to a select group of global security leaders dedicated to protecting essential infrastructure and sharing findings and best practices across industries. The release quotes Verizon CEO Dan Schulman: "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." The release also includes a longer Schulman quote noting that Verizon has been "rigorously testing" the technology and calling Verizon the "only telecommunications company utilizing Mythos Preview."

Generative models with advanced code- and system-reasoning capabilities, like those described by Anthropic for Claude Mythos Preview, are increasingly being evaluated for vulnerability discovery, threat-surface mapping, and triage. Industry-pattern observations: organizations that pilot such models typically use them for fast pattern matching across large codebases, automating common static-analysis tasks, and surfacing candidate exploits that then require human validation. These workflows can raise trade-offs between faster discovery, false positives, and the operational burden of validation and safe handling of model outputs.

Industry context: Project Glasswing follows public reporting and vendor announcements this year about using frontier LLMs in cybersecurity roles, and it positions Anthropic among vendors offering model access for defensive security use cases. For practitioners, participation by a major telecom like Verizon signals growing enterprise appetite to combine proprietary networks and operational knowledge with external model capabilities under controlled access. Reporting by Seeking Alpha and MarketScreener reiterates the Verizon release and frames the move within competitive dynamics between AI vendors pursuing security products.

• •Adoption signals: announcements of additional utility members or cross-industry findings published by Project Glasswing participants.
• •Operational outputs: whether participating organizations publish reproducible detection patterns, vetted vulnerability disclosures, or tooling integrations that can be evaluated by security teams.
• •Safety and governance: the access controls, red-team results, and usage policies Anthropic and participating organizations publish around Claude Mythos Preview to manage misuse and accidental disclosure.

For practitioners: monitor whether pilot results translate into repeatable, low-noise detection that integrates with existing SAST/DAST pipelines and vulnerability management workflows, and watch public documentation from Project Glasswing for governance and reproducibility details.