SourceHut's Git Service Disrupted by LLM Crawler Botnets
Key Takeaways
- ▸LLM crawler botnets have successfully bypassed SourceHut's existing security defenses
- ▸The aggressive scraping is causing severe performance degradation for git.sr.ht users
- ▸This is part of an ongoing pattern of LLM-related DDoS attacks against SourceHut's services
Summary
SourceHut's git.sr.ht service is experiencing significant performance degradation due to an aggressive crawler botnet that is scraping the platform to gather training data for large language models. The botnet has successfully circumvented SourceHut's existing defense mechanisms, causing severe disruptions to the git service's web frontend. System administrators are actively investigating the incident to restore normal operations.
This represents an ongoing challenge for the code hosting platform, which has faced repeated DDoS attacks from LLM crawlers in recent weeks. The attacks highlight a growing tension between AI companies training large language models on publicly available code and the infrastructure costs imposed on open-source platforms that host that code.
- The incident underscores infrastructure challenges posed by indiscriminate AI training data collection
