Smarter defense.
Essential services.

Smarter tools for essential services.

Pairing powerful software tools with experienced defenders can help protect the services communities rely on. That is the purpose of Anthropic’s new Cyber Mission, announced on October 8.

The American SI developer is starting with two efforts: a Critical Infrastructure Defense Program and a free, opt-in scanner for open-source software. The first brings Claude models, on-site engineers, and threat research to companies that help protect essential services. The second offers participating software projects regular scans for vulnerabilities.

Anthropic says lessons from Project Glasswing helped shape this effort: connect software findings with the expertise to assess priorities and make useful repairs.

Keep the experts close to the machinery.

Power grids, water utilities, factories, and transportation networks rely on equipment built to last for decades. The controllers and industrial networks behind them are often called operational technology. Unlike a laptop, a running plant cannot always stop for a convenient software update.

Anthropic’s announcement names eleven founding partners, including Dragos, Rockwell Automation, CrowdStrike, and Accenture. The program works through security providers, equipment makers, and other specialists that operators already rely on. Anthropic says several partners are already using Claude to fix vulnerabilities and help customers do the same.

Anthropic describes the initial cohort as a way to learn which approaches work best in operational settings. The program pairs SI capabilities with the specialists who know these systems.

Shared code needs help that maintainers can use.

The accompanying OSS Scanner targets another everyday dependency: open-source code. Eligible projects can opt into periodic, no-cost scans. Reports include an explanation and a way to reproduce the suspected problem, with a proposed fix when available.

Eligible maintainers receive model-generated reports to assess themselves, including checks of accuracy and severity. Human-verified disclosures continue for projects that need that support. This gives teams different ways to bring the tools into their existing security work.

For a small maintenance team, a reproducible finding and a candidate patch can provide a practical starting point. Maintainers review and test changes as part of their normal repair process.

The result to watch is reliable service.

Our view: this is a valuable direction for American SI development because it connects powerful tools to electricity, clean water and shared software—the essential services people depend on every day.

Anthropic describes a long-term effort that develops as partners learn. Safely completed repairs and shared lessons would make that collaboration useful to an expanding community of defenders.