Google Pauses Open-Source Bug Bounty as AI-Generated Reports Overwhelm Maintainers
Google Pauses Open-Source Bug Bounty as AI-Generated Reports Overwhelm Maintainers
Google has suspended new submissions to its Open Source Software Vulnerability Reward Program, known as the OSS VRP, effective October 1, 2026. The company says the pause was prompted by an overwhelming influx of low-quality vulnerability reports, many of which appear to be AI-generated and, in a number of cases, outright hallucinated. Many observers note that the sheer volume of these submissions had begun consuming more maintainer time than it was saving.
Google Hits Pause on Its Open-Source Bug Bounty
The suspension specifically targets product vulnerability submissions to the OSS VRP. Reports filed before October 1 are continuing to be processed as normal, and the pause does not extend to OSS VRP supply chain reports or to certain Google Cloud VRP repositories, which remain open for submissions. The scope of the freeze suggests Google is trying to isolate the part of the program most affected by the reported surge, rather than shutting down vulnerability reporting altogether.
The AI Slop Problem Behind the Decision
According to Google, a surge of low-effort, superficially plausible but substantively invalid reports is at the center of the decision. A recurring concern raised across coverage is that these submissions often carry the hallmarks of AI-generated content: reports that read as coherent and technically formatted, yet describe vulnerabilities that do not actually exist. Reviewing and ruling out these reports takes real human time, time that maintainers say would otherwise go toward investigating genuine security issues.
This is not the first attempt to address the problem. Google reportedly adjusted its VRP criteria back in March 2026 in hopes of filtering out lower-quality submissions. That adjustment does not appear to have been sufficient on its own, which industry observers frame as evidence that the issue has continued to grow rather than level off.
Not Just Google: An Industry-Wide Strain
Coverage of the suspension frames it as part of a broader pattern rather than an isolated incident. Linux kernel maintainers have reportedly seen a record number of vulnerability submissions per release cycle, a volume widely attributed to AI-powered bug-hunting tools. Intel has also paused its own bug bounty program, though Intel has not officially confirmed that AI-generated reports were the cause, and that connection should be treated as speculation rather than a confirmed fact at this stage.
Other programs, including the Internet Bug Bounty, are also described as facing similar pressures. Taken together, these cases suggest an industry-wide strain on vulnerability reporting systems as generative AI tools make it easier to produce reports that look credible on the surface, whether or not they hold up under scrutiny.
The Human Cost for Open-Source Maintainers
A recurring theme in discussion of this episode is the position of open-source maintainers themselves. Many are volunteers or operate with limited institutional support, and triaging a growing stream of invalid automated reports diverts time and attention away from other priorities, including genuine security work. This has prompted broader questions about whether bounty programs as currently structured can remain sustainable as AI tooling continues to scale the volume of submissions, regardless of their quality.
The situation also highlights a tension underlying the original purpose of bug bounty programs: lowering the barrier to entry for security research is generally seen as valuable, but when that barrier drops low enough for automated or semi-automated tools to flood review pipelines, the programs can end up working against the maintainers they were designed to support.
What Comes Next
Google has indicated that a formal update on potential reforms to the OSS VRP is expected in the first quarter of 2027. In the meantime, it remains unclear exactly how long the current suspension will last or what changes might eventually be introduced. Legitimate security researchers have, for now, fewer avenues to submit and be rewarded for genuine findings within the affected part of the program.
More broadly, the episode points to an open question for the security research community: how to design verification mechanisms that can distinguish authentic vulnerability reports from AI-generated noise, without reintroducing the kind of friction that bounty programs were originally created to remove.