It's remarkably easy to inject new medical misinformation into LLMs

3 pointsby prng20214 months ago

1 comment

gnabgib4 months ago

The paper: [Medical large language models are vulnerable to data-poisoning attacks](<a href="https://news.ycombinator.com/item?id=42640260">https://news.ycombinator.com/item?id=42640260</a>)