Anthropic has introduced a free scanning service, OSS Scanner, for regularly searching for vulnerabilities in open source projects. The service uses the developer’s most powerful AI models, including Claude Mythos.
Image source: Anthropic
Projects will be able to connect to the service voluntarily using an opt-in subscription, which provides for the user’s voluntary consent to receive notifications and reports about found vulnerabilities. Anthropic expects that automated scanning will identify problems faster and earlier than using traditional screening methods alone.
However, OSS Scanner reports will not be reviewed by specialists. Anthropic emphasizes that all results are generated by AI without human participation in the verification and initial assessment of detected problems. Therefore, messages may contain erroneous or invalid findings that developers will have to check themselves.
OSS Scanner comes amid growing interest in using AI to find software vulnerabilities. In recent months, such tools have helped uncover several serious problems in open source projects, including the critical Copy Fail vulnerability that affected almost all Linux distributions in May.
At the same time, an increase in the number of automatic checks creates additional burden on developers. Some open source projects are already struggling to cope with the influx of bug reports generated by AI models. Among those who encountered this problem were Linus Torvalds and even Google.
If you notice an error, select it with the mouse and press CTRL+ENTER.
