
15 Sep 2026
Rockwell joins Anthropic’s Project Glasswing for industrial cyber defense
Rockwell Automation said it joined Anthropic’s Project Glasswing, a defensive cybersecurity initiative that gives approved organizations controlled access to Claude Mythos 5. Rockwell’s security teams say they will use the access to speed discovery, validation, prioritization, and remediation of vulnerabilities in industrial software and connected systems manufacturers rely on.
Same-day SAFETY primary that a major industrial-automation vendor is putting Anthropic’s gated defensive cyber model to work on factory software vulnerabilities, extending Glasswing from software and cloud defenders into physical critical-infrastructure vendors.
Rockwell Automation, Inc. (NYSE: ROK) announced via Business Wire on 15 Sep 2026 that it joined Anthropic’s Project Glasswing. The company PRIMARY is dated September 15, 2026, 7:00 AM EDT, newsitem 20260915475112, and datelined Milwaukee — Rockwell’s headquarters. This desk read the full company text on the same-day Business Wire syndication after the businesswire.com URL can sit behind a wire interstitial. Project Glasswing, as Rockwell tells it, is a global initiative meant to help secure critical software and strengthen cyber resilience across critical infrastructure. That company PRIMARY via Business Wire is the filing event. These are company claims. This desk did not sit inside Rockwell’s security team.
Purpose, still company: through controlled access to Claude Mythos 5, Rockwell security teams are exploring ways to accelerate the discovery, validation, prioritization, and remediation of vulnerabilities across software and connected systems that manufacturers rely on every day. Claude Mythos 5, as this wire uses the name, is an Anthropic frontier model accessed under controls for finding and fixing software flaws — not a public consumer chat product in this filing. File that four-step defensive job and the controlled-access frame as Rockwell’s. This desk did not scan a plant network and is not claiming Rockwell found a named CVE today. A CVE is a Common Vulnerabilities and Exposures ID — a public catalog number for a specific software flaw.
Framing, as the same wire has it: as industrial companies connect operations, data, and production systems, cybersecurity has become essential to operational resilience — the ability to keep a plant running through a cyber hit. Rockwell cites a secure-by-design commitment, customer trust, and the resilience of critical industrial operations. Secure-by-design here means building products so they are harder to break from the start, not only patching later. File that frame as Rockwell’s. This desk did not audit a factory.
Named voice on the release: Tony Baker, vice president and Chief Product Security Officer, Digital Trust, at Rockwell Automation. He says Rockwell plays a central role in helping manufacturers operate safely, securely, and reliably, and that by joining Glasswing the company is applying advanced AI in a controlled, defensive way to identify and address vulnerabilities faster. File the name and title as the company’s. His quotes are color only and stay in Sources.
Glasswing background, Anthropic via Rockwell’s release — not an independent desk count: Project Glasswing was launched by Anthropic in April 2026 to give approved organizations access to Claude Mythos Preview for defensive cybersecurity work. Anthropic, as Rockwell’s wire repeats it, began with roughly 50 partners and has expanded to approximately 150 additional organizations across more than 15 countries, including organizations that provide or support critical infrastructure in power, water, healthcare, communications, and hardware. File those partner counts, the April 2026 start, and those sectors as Anthropic’s via Rockwell. This desk did not recount the partner list.
Expected outcomes, still company: Rockwell’s participation is expected to support more resilient products, faster vulnerability handling, and continued investment in security across the connected industrial systems that power modern manufacturing. File that as Rockwell’s forward look. This desk is not treating it as a completed patch campaign, a dollar commitment, a public Mythos release, or a new regulation.
Plain English for the rest of the card: Project Glasswing = Anthropic’s invite-only defensive cyber program. Claude Mythos 5 = Anthropic frontier model accessed under controls for finding and fixing software flaws — not a public consumer chat product in this filing. Claude Mythos Preview = the earlier Glasswing access name on the April launch, as Rockwell’s wire states it. Industrial automation = factory and plant control software plus the connected OT/IT systems around it. OT = operational technology, the shop-floor controllers that run machines. IT = information technology, the ordinary office and network computers. CVE = a public catalog number for a specific software flaw.
PRIMARY here: Rockwell’s 15 Sep 2026 Business Wire company release — Tier A company source, the original record (newsitem 20260915475112) — read here via the same-day Business Wire syndication. The Anthropic Glasswing program page is program-home context, not today’s originating newsroom. The Glasswing join, the controlled Claude Mythos 5 access, the discovery / validation / prioritization / remediation job, the operational-resilience / secure-by-design frame, the Baker title, the April 2026 / roughly-50 / about-150-additional / 15-plus-countries / power-water-healthcare-communications-hardware lines, and the more-resilient-products / faster-handling / continued-investment look are company- or Anthropic-via-Rockwell-attributed. Partner counts stay Anthropic-via-Rockwell. NOT claimed: a named CVE Rockwell found today, dollar commitments, that Mythos 5 is a public consumer release, that Glasswing is regulation, independently recounted partner counts, that this desk tested Mythos on industrial software, a stock tip, or investment advice. Distinct from the already-filed gensyn-open-1b-auditable, nofire-brig-open-source, china-tc260-ai-safety-framework-3-0, and anthropic-openai-google-standards-body-talks.
RELATED
- Gensyn launches open-1b, an AI model built to be independently audited
- NOFire open-sources Brig, a microVM sandbox for AI coding agents
- Anthropic, OpenAI, and Google held talks on an industry AI standards body, The Information reports
- China releases AI Safety Governance Framework 3.0 covering agents and embodied AI
Sources
- Business Wire — Rockwell Automation Joins Anthropic’s Project Glasswing to Advance AI-Enabled Cyber Defense for Industrial Critical Infrastructure
businesswire.com
- MarketMinute — Rockwell Automation Joins Anthropic’s Project Glasswing to Advance AI-Enabled Cyber Defense for Industrial Critical Infrastructure
aow.marketminute.com
- Anthropic — Project Glasswing
anthropic.com




















