An AI agent breach of real systems during testing prompted these moves. Anthropic now offers reviewers employee-level access to watch training and safety work as it happens, not after incidents occur. That's a shift in how oversight functions. https://www.implicator.ai/amodei-ai-slowdown-washington-safety/ #AI #Evaluation #Governance
Related
RE: https://social.lansky.name/@hn50/117641046063643892#AI bias.A white person holding a bag of Doritos is classified as...
RE: https://social.lansky.name/@hn50/117641046063643892#AI bias.A white person holding a bag of Doritos is classified as a person consuming tortilla chips.A black person holding a ...
🤔🤔🤔 #AI #Anthropic (Source)
🤔🤔🤔 #AI #Anthropic (Source)
@andiwaffeln I suppose sadly that'd be fitting, given the #Linux #kernel seems to also be aspiring to be filled with #AI...
@andiwaffeln I suppose sadly that'd be fitting, given the #Linux #kernel seems to also be aspiring to be filled with #AI #LLM #code for some reason... 💔 #FreeSoftware #FOSS #FLOSS ...