Anthropic launches free OSS Scanner and critical infrastructure cyber program

The Critical Infrastructure Defense Program begins with 11 founding partners. Accenture, Booz Allen Hamilton, Deloitte & Touche and PricewaterhouseCoopers come from the consulting side, where they run security programs for operators. Security vendors make up the largest group, with industrial specialists Dragos, Insane Cyber and Nozomi Networks joining CrowdStrike and Palo Alto Networks. Hitachi and Rockwell Automation build and patch the hardware itself.

The program targets operational technology in plants, substations and water systems, where equipment built to run for decades often cannot be taken offline for a patch, so known flaws can remain for years. Anthropic said several partners are already using Claude to fix vulnerabilities and to help their customers do the same. More partners and sectors are due to join in coming months.

Andrew Turner, president of commercial cyber at Booz Allen, called operational technology "the next frontier for autonomous AI-enabled attacks" in comments published with the announcement. He said what matters now is how much control artificial intelligence can gain over an industrial process and how fast.

The announcement leaves out commercial terms. Axios noted in its coverage that Anthropic has not said whether partners get free model access or who covers computing costs. Axios also questioned how partners will test and deploy fixes without disrupting utility operations.

OSS Scanner, launched with the infrastructure program, offers open-source projects free periodic vulnerability scans from Anthropic's strongest models. Anthropic said projects that opt in will get thorough, periodic security scans at no cost. The service was inspired by Google's OSS-Fuzz, which runs fuzzers against open-source code.

The open-source service grew out of a backlog in Anthropic's own disclosure work. Of more than 29,000 candidate vulnerabilities its models turned up in widely used software over the past six months, staff had manually reviewed about 6,000. A growing number of maintainers who received early reports asked for the whole unreviewed batch, including proposed patches, and nearly 5,000 such reports have gone out so far, according to SiliconANGLE.

Every OSS Scanner report includes a self-contained reproducer. Where the model can manage it, a candidate patch comes with the write-up, and a bisection of the code history shows when the flaw was introduced. Anton Arapov of the OpenSSL Corporation, an early tester, said a report with a real exploit attached is "basically job done for an engineer."

Reports from the new service go straight to maintainers without human review so they arrive faster. Anthropic said some will contain mistakes as a result, such as a wrong severity rating. The outputs are fully model-generated, without human review or triage, which enables faster and more frequent scanning but means reports may be incorrect or invalid, according to The Verge.

Before opening the scanner to more projects, Anthropic tested its accuracy. Expert penetration testers who vet the company's coordinated disclosures checked 97 critical and high-severity findings from an early version across 48 projects and cleared 85 for disclosure. Of the other 12, all but one were real bugs that duplicated known issues or other findings from the scan. Embedded encryption library developer wolfSSL said all but two of the 74 reports it received during early trials were valid and that five became CVEs.

Core maintainers can enroll by submitting a pull request to an Anthropic GitHub repository. Eligibility follows the OSS-Fuzz test of "critical impact on infrastructure and user security," with decisions made case by case. Projects without the staff to keep up with raw findings will still get human-verified reports through Anthropic's existing disclosure process.

The Defender Advantage Fund that Anthropic set up in August pays to keep the scanner free. Anthropic has also put money into the Python Software Foundation and the Apache Software Foundation, as well as Alpha-Omega and OpenSSF through the Linux Foundation.

Both launches draw on lessons from Project Glasswing, which gave vetted organizations access to Claude Mythos from April until it was folded into an expanded Cyber Verification Program earlier this week. Finding vulnerabilities has never been easier, according to Anthropic, but verifying, prioritizing and fixing them remains hard. The company said Glasswing has not yet cut cyber risk by enough, and it expects AI to favor defenders within two years.

OSS Scanner is not the first AI bug-hunting helper. AI tools have helped find major security flaws in open-source software in recent months, such as the "Copy Fail" bug that affected nearly every Linux distribution in May, according to The Verge. At the same time, some open-source projects are struggling to keep up with the sudden onslaught of AI-generated bug reports, including Linus Torvalds and Google.