AutoTrust's Hugging Face Models Accused of Using Bot Networks for Trending Status

An investigation revealed that machine learning models from the organization AutoTrust were artificially boosted to the Hugging Face trending page using thousands of AI-generated fake accounts and farmed likes. The analysis highlighted clear patterns among the accounts, including random name generator usernames, AI-generated biographies, and coordinated liking behavior. This incident exposes critical platform manipulation vulnerabilities on Hugging Face, highlighting how bad actors can game recommendation algorithms to gain artificial visibility and authority. Such security issues can mislead developers into trusting compromised or low-quality models for their AI pipelines. The investigation compared AutoTrust's metrics against established organizations like Google and Meta, finding suspicious join-date distributions and automated interactions. Running such a bot network requires bypassing security measures like hCaptcha and AWS WAF, indicating a sophisticated and costly effort.