AI Company Caught Faking Thousands of Likes on Hugging Face

If you've ever browsed Hugging Face, a website where people share AI models, you might have seen AutoTrust's models trending at the top. They claimed impressive results, but a Reddit user decided to dig deeper. What they found was shocking: thousands of likes on AutoTrust's models came from fake accounts. These accounts had realistic names, AI-written bios, and even liked a few random papers to look real. But they were all created just to boost AutoTrust's popularity.

The Reddit user analyzed likes on models from reputable companies like Google and Meta and compared them to AutoTrust's. The difference was huge. AutoTrust's models had a suspiciously high number of likes from accounts that were created recently and had no other activity. Another user, prithivMLmods, looked at AutoTrust's spaces (interactive demos) and found the same pattern. The fake accounts were likely created in bulk, and the cost to generate them—including solving captchas—isn't cheap, suggesting a deliberate and coordinated effort.

So why should you care? When you're looking for an AI tool, you might rely on popularity signals like likes or trending status to decide what's good. If those signals are faked, you could end up using a model that doesn't work well or, worse, one that's designed to scam you. This incident shows that even on trusted platforms, you need to be skeptical of sudden popularity and do your own research.

Hugging Face hasn't commented yet, and AutoTrust hasn't responded to the allegations. But the evidence is strong. This isn't the first time someone has tried to game the system, and it's a reminder that online popularity can be bought. As AI becomes more integrated into our lives, it's more important than ever to verify what you're using.