- cross-posted to:
- [email protected]
- cross-posted to:
- [email protected]
Website operators are being asked to feed LLM crawlers poisoned data by a project called Poison Fountain.
The project page links to URLs which provide a practically endless stream of poisoned training data. They have determined that this approach is very effective at ultimately sabotaging the quality and accuracy of AI which has been trained on it.
Small quantities of poisoned training data can significantly damage a language model.
The page also gives suggestions on how to put the provided resources to use.
As someone who makes and uses software, I feel it is not okay to steal source code. I wouldn’t feel okay with myself getting something for free when it’s based on the stolen work of tens of thousands of people
AI companies aren’t respecting crawler blocking. They’re actively working to ensure their crawlers bypass any anti-crawler protections
As a side note, these efforts help AI in the long-term. If we can poison LLMs, then you can guarantee a state actor can as well. AI needs to be able to weather training data attacks, otherwise they become an easily manipulated propaganda tool