GitHub is reevaluating its approach to pull requests as the influx of low-quality contributions, particularly those generated by AI, becomes a significant issue for the open source community. Product manager Camilla Moraes highlighted that maintainers are spending excessive time sifting through submissions that often fail to meet project standards. These contributions frequently do not adhere to guidelines, are abandoned shortly after submission, and are often AI-generated. Moraes opened a community discussion to solicit input on possible solutions, including options to disable pull requests or restrict them to project collaborators. Concerns about AI-generated code are echoed by several industry figures. Xavier Portilla Edo from Voiceflow noted that only about 10% of AI-created pull requests meet quality standards. Developers like Daniel Stenberg have experienced the burden of low-quality AI contributions firsthand, leading projects like curl to shut down their bug bounty programs to discourage submissions that lack quality. Other developers, such as Jiaxiao Zhou, expressed worries about the broken trust model in code reviews, where maintainers can no longer assume that authors understand the code they submit. They face increased cognitive loads, needing to evaluate both the code and the author's comprehension. The discussion has also raised issues about community incentives in open source. Nathan Brake from Mozilla.ai emphasized the risk of knowledge sharing diminishing as AI does more of the coding work. Chad Wilson shared his frustrations with poorly constructed pull requests, pointing out that without clear disclosure regarding AI involvement, trust within the community could erode. GitHub's Matthew Isabel responded by stating the focus should be on the quality of pull requests rather than the authorship. He acknowledged that the volume of submissions has surged due to AI, stressing the need for new tools and workflows to manage the increasing workload on maintainers.