Stefan Eissing on Nostr: Red-Teaming AIs: „Due to fundamental limitations of language models, one must ...

Red-Teaming AIs: „Due to fundamental limitations of language models, one must assume that if an LLM is supplied with untrusted input, it will produce arbitrary output.“
🙃

https://arxiv.org/abs/2501.07238 (via nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqf6a33pfyp67y8llhunlhrf855xm47n3fdqymvxfj7yx78c6vqf4sy8ssyg (nprofile…ssyg))