22 Mar
2023
22 Mar
'23
2:20 p.m.
the article prof rat linked for me opens to a humanoid robot. i found a cracking-with-chatgpt article on what prof rat quoted at https://www.theregister.com/2023/02/22/chatgpt_pwn2own_ai/ . i selected this link because the reporter interviewed the participants rather than rehashing another newsfeed. the articles seem to focus on the participants using chatgpt for server development and stress the service is not sufficient for e.g. vulnerability identification . it seems to me with techniques like those used in langchain or llama-index that vulnerability identification and automated exploitation would be wholly doable and are likely already being done.