If your organisation depends on open-source software — and virtually every organisation does — a landmark new programme from Anthropic deserves your full attention. On October 8, 2026, Anthropic unveiled OSS Scanner as part of its broader Anthropic Cyber Mission, making AI-powered vulnerability scanning freely available to eligible open-source projects. For security teams, this is not just another vendor announcement. It signals a fundamental shift in how the open-source ecosystem can be defended — and it changes what you should expect from the libraries, frameworks, and tools your business runs on.

What Is Anthropic's OSS Scanner?

Anthropic has introduced OSS Scanner, a free service that runs automated security checks on open-source software projects using its Claude models.

Maintainers who opt in receive what the company describes as thorough, periodic scans designed to flag vulnerabilities at no cost.

Inspired by Google's OSS-Fuzz, OSS Scanner is an opt-in service where enrolled projects receive periodic scans from Anthropic's most capable models, free of charge.

Each report includes a proof of concept of how the bug could be exploited, an explanation, and a suggested fix where one is available — and the reports are model-generated and sent without human review.

Crucially,

while Claude Security, Anthropic's general-access code scanning and patching product, focuses on helping enterprises defend their systems, OSS Scanner provides security audits at no cost to open-source projects.

The Backstory: Project Glasswing and the Scale of the Problem

OSS Scanner didn't emerge from thin air. It grew directly from Anthropic's internal research programme called Project Glasswing — and the numbers behind it are sobering.

Over the past six months, Anthropic tested its latest AI models on some of the most widely used open-source software and found more than 29,000 possible vulnerabilities. Its experts have had time to manually check and confirm only about 6,000 of them — a large number of real bugs, but the backlog shows that the main limit is not the AI's ability to find flaws.

Anthropic has already shared nearly 5,000 unverified reports, along with suggested fixes, with maintainers who requested them. When attackers can develop an exploit in minutes, waiting for every report to be checked by a human before sharing it may create more risk than sending it early, as long as maintainers understand that the findings still need verification.

This tension between speed and verification is central to understanding how OSS Scanner works — and what security teams need to be aware of.

The Broader Anthropic Cyber Mission

OSS Scanner is just one component of a significantly larger effort.

Anthropic launched the Anthropic Cyber Mission, a long-term initiative to help secure critical infrastructure and open-source software by providing defenders with AI tools, research, and technical resources.

The initiative covers critical infrastructure, starting with securing the operational technology behind power grids, water systems, and transportation networks. Anthropic introduced the Critical Infrastructure Defense Program (CIDP), which brings frontier models, on-site engineers, and threat research to the defenders that protect operational technology.

Eleven companies, including Hitachi, joined as founding partners of CIDP.

The initiative pairs Claude models and on-site engineers with security providers like CrowdStrike, Dragos, and Accenture to secure operational technology.

For security teams in enterprise environments, the significance is clear: Anthropic is positioning AI not as a threat to manage, but as active infrastructure for defence.

How AI Vulnerability Detection Has Improved Dramatically

Sceptics might wonder whether AI-generated security reports are actually good enough to act on. The data says they are — and are improving rapidly.

On the CyberGym benchmark, AI models went from finding fewer than 20% of vulnerabilities at the beginning of last year to more than 85% this year.

Anthropic's free AI OSS Scanner uses Claude Mythos to find open-source flaws, and tests saw 85 of 97 high-severity findings pass validation.

More precisely,

of 97 critical/high-severity reports reviewed, 85 (88%) met coordinated disclosure standards, 11 duplicated known bugs, and only one was invalid.

Separately from that tester check, Anthropic says it expects a true-positive rate above 90% for the service and will work to improve it over time.

These are metrics that would be competitive with traditional human-led penetration testing at a fraction of the cost and time — making OSS Scanner a genuinely useful signal for security teams, provided they treat reports as requiring validation rather than as final verdicts.

How to Enrol an Open-Source Project

If you maintain or contribute to a qualifying open-source project, the enrolment process is deliberately lightweight.

Core maintainers enrol by opening a pull request to the anthropics/oss-scanner GitHub repository that adds a project.yaml config file. The file includes a repository link, a contact address, and a path to a Dockerfile, so the scanning agent can build and audit the project in a sandbox with no internet access.

Eligibility, however, is not universal.

Eligibility isn't open to everyone — Anthropic borrows the OSS-Fuzz test of "critical impact on infrastructure and user security" and decides case by case. It will confirm that an applicant is a core maintainer first and may tighten the criteria depending on demand.

Anthropic weighs projects' exposure to remote attacks and how many users or projects depend on them, evaluating requests case by case and manually checking that applicants are core maintainers.

Regarding disclosure timelines,

the reports skip human review, there is no 90-day public disclosure clock, and Anthropic pays the compute bill.

Anthropic recommends that the contact email be an address suitable for publication, such as a security alias.

What Security Teams Need to Consider

For enterprise security and AppSec teams, OSS Scanner raises several critical operational questions worth addressing now.

Understand the No-Human-Review Model

Unlike Anthropic's existing coordinated vulnerability disclosure process, reports from OSS Scanner are fully generated by AI without human review or triage. That means maintainers receive findings faster, but Anthropic warns that some reports can still be incorrect.

Anthropic flags that reports can be wrong, with severity ratings that "can be inflated or the scanner misunderstood the project's threat model."

Security teams consuming these reports downstream must factor this into their triage workflows.

Know the Competitive Context

From a market perspective, Anthropic's entry intensifies competition in the developer-focused security space, where incumbents like Snyk, Veracode, and GitHub Advanced Security have built paid models around human-reviewed findings. If OSS Scanner gains traction, it may force rivals to reconsider pricing or to incorporate more AI-driven components, reshaping the economics of SaaS security tooling.

Recognise What OSS Scanner Doesn't Replace

OSS Scanner does not scan every dependency used by every business, and enrollment is limited to projects Anthropic accepts. It also does not remove the need for existing dependency monitoring, patch management, and internal security testing.

Practical Tips: What Security Teams Should Do Right Now

Here is a concrete action list based on what we know about OSS Scanner and the Anthropic Cyber Mission:

OSS Scanner emails maintainers vulnerability reports with a proof of concept and a suggested fix. You enrol by opening one pull request on GitHub with a small YAML file and a Dockerfile.

The barrier to entry is low for eligible projects.

Maintainers can supply a threat model to explain the intended security boundaries and severity rules

, which helps the scanner contextualise findings more accurately.

The expanded Cyber Verification Program separates access into Defense, Red Team, and Specialized tiers, with verification requirements and safeguards matched to authorised activities. Specialised Access covers sensitive systems where testing could affect physical safety or disrupt essential services.

Enterprise security teams should understand which tier applies to their context.

Conclusion: A New Era for Open-Source Security

Open-source software underpins much of the world's digital infrastructure, yet many projects are maintained by small volunteer teams with limited capacity for rigorous security review. By offering AI-driven scanning at no charge, Anthropic is effectively extending enterprise-grade security tooling to a segment of the software ecosystem that has historically lacked the resources to access it.

For organisations pursuing digital transformation, the move signals a broader shift: AI models are increasingly being deployed not just to generate code but to audit and secure it. This changes the operating model for software risk management, embedding automated oversight earlier and more continuously in the development lifecycle rather than relying solely on periodic human audits.

OSS Scanner is not a silver bullet — but it is a meaningful, well-validated leap forward for open-source security. The question for security leaders is not whether to pay attention to this programme, but how quickly they can integrate awareness of it into their supply chain security strategy.

Ready to strengthen your organisation's open-source security posture? Start by reviewing your dependency inventory today, flag any maintainer relationships your team holds, and explore enrolment at Anthropic's official OSS Scanner GitHub repository. The future of software supply chain security is AI-assisted, continuous, and — for qualifying projects — now free.