I agree with this in spirit, but we use this language for inanimate objects that lack agency all the time. If I built a car and let it roll down a hill with no one in it and it hit a person, we would say “the car ran a person over”. This does not imply agency on the car’s part, and it does not imply that I am not responsible.
Liability should always fall back to the whoever is running the bot.
That said, a public-facing AI with access to internal data is itself a security risk. People bully and cajole AI’s into disclosing private information and sometimes it is laughably easy to do so.
People working at Meta will end up leaking those info. Muse is just the tool.
Saying AI is gonna do this bad thing etc is just offloading liability
This kind of language is frustrating and should be changed because it is people who make these things and they should be held responsible.
B–B–Bingo.
I hate how these stories set us up to blame the product and not the shitty vibe coders.
I agree with this in spirit, but we use this language for inanimate objects that lack agency all the time. If I built a car and let it roll down a hill with no one in it and it hit a person, we would say “the car ran a person over”. This does not imply agency on the car’s part, and it does not imply that I am not responsible.
Be that as it may, it’s very important to keep the shitty developers in these conversations. Most people don’t make that logical leap automatically.
Liability should always fall back to the whoever is running the bot.
That said, a public-facing AI with access to internal data is itself a security risk. People bully and cajole AI’s into disclosing private information and sometimes it is laughably easy to do so.