Rockwell Brings Expertise to Anthropic’s Cybersecurity Project

Rockwell Automation has joined a pioneering, collaborative, cyber-defense-focused initiative launched by AI company Anthropic. By joining Project Glasswing and undergoing a rigorous vetting process, Rockwell gains access to Anthropic’s Mythos 5 AI model. Rockwell will leverage AI's untapped capabilities to identify key vulnerabilities in software and industrial automation systems, supporting the development of more resilient products and solutions that help safeguard industrial systems powering modern manufacturing.

As industrial automation products and solutions evolve and manufacturing facilities collect more data, ensuring data integrity and the security of enterprise–wide industrial systems has never been more important. Image used courtesy of Pixabay

Cyber Threats are on the Rise

This collaboration brings cutting-edge AI to defending global manufacturing infrastructure, with Rockwell supplying the hardware, software, and industrial control systems (ICS) that drive modern manufacturing facilities, utilities, and supply chains.

Manufacturers face growing cyber threats as they digitize and integrate IT networks with operational technology (OT) on plant floors. Flat networks enable lateral movement, outdated legacy systems lack native protection, and costly downtime makes routine patching difficult, driving these risks. Claude Mythos 5 uses deep vulnerability detection, rigorous simulation, intelligent alert prioritization, and human-in-the-loop patching and fixes to help manage these complex challenges.

Project Glasswing

Project Glasswing is a global cybersecurity defense initiative that brings together leading players across cybersecurity, cloud, silicon, and global digital infrastructure to identify software vulnerabilities across key infrastructure and strengthen defenses against machine-speed cyber threats.

Founding launch participants include Cisco, NVIDIA, Google, Amazon Web Services, Anthropic (as initiative organizer), Apple, Broadcom, CrowdStrike, JPMorganChase, the Linux Foundation, Microsoft, and Palo Alto Networks. Initially, participants received an unreleased frontier AI model to test its ability to pinpoint software vulnerabilities. Following successful testing, Anthropic developed and launched its “Mythos class” large language model, Claude Mythos 5. Mythos 5 is not for public use, and partners joining the Project Glasswing initiative must undergo strict vetting before receiving authorization to use it.

Since the initiative launched, Anthropic has opened access to Claude Mythos 5 through Project Glasswing to 40+ organizations, which has expanded to around 150 spanning 15 countries worldwide.

An overview of Claude Preview, Mythos 5, and Project Glasswing. Video used courtesy of Anthropic

Anthropic offers several different models: Haiku, Sonnet, Opus, and Fable. Mythos is a generational upgrade intended for deep domain investigation, autonomous agentic processes, and sophisticated multi-step reasoning. In domains like software development and molecular biology, these models exhibit previously unheard-of ability in complicated reasoning and long-horizon problem-solving.

Shoring up the Industrial Ecosystem

This collaboration brings cutting-edge AI to defending manufacturing infrastructure worldwide, with Rockwell supplying its technology and software expertise.

Rockwell Automation has joined Project Glasswing’s collaborative league of cyber defenders and partners to help secure software and systems within and across critical industrial infrastructure. Rockwell’s participation in this global initiative reinforces its commitment to delivering top-tier secure–by–design products and solutions to its broad customer base. The goal is to create more resilient industrial infrastructure that can withstand the continued growth of cyber threats as automation systems evolve.