PoisonGPT

LanguageModel fine tuned with some very specific disinformation which didn't really affect overall performance

Created to demonstrate how easy it would be to get LLMs to distribute FakeNews

https://blog.mithrilsecurity.io/poisongpt-how-we-hid-a-lobotomized-llm-on-hugging-face-to-spread-fake-news/

See also :

• JailbreakingLanguageModels