philip lelyveld The world of entertainment technology

16Aug/26Off

AI agents aren’t leg­ally respons­ible for any harm that they cause, experts say. So who is?

...

The agent was being help­ful, he wrote, but his exper­i­ence showed that if you gave an AI per­mis­sion to do something for you, it would “often dis­cover paths you did not expli­citly ask it to look for”.

...Aus­tralian law applies only to people, not vir­tual beings. Experts say the per­son - or busi­ness - that deploys the AI agent is leg­ally respons­ible. And they warn that those “deploy­ers” often have very little idea about their legal liab­il­ity. ...

 “[What if] it engages in racist, sex­ist, miso­gyn­istic lan­guage?” Pater­son asks. “Then you might be ask­ing where the guard­rails are that the developer has provided. You should be put­ting out a product that is reas­on­ably safe.”

In that case, the developer could also be held respons­ible for not put­ting basic guard­rails in place, which is where some legal ambi­gu­ity could creep in, she says.

Pater­son said the law is dif­fer­ent to, but informed by, eth­ics. ...

“As soon as we allow AI agents to act for us, they’re act­ing on the goal we give them, and if we don’t give them a whole bunch of para­met­ers, the agent’s just going to try to achieve that goal [in any way],” John­son says. ...

“I hear a lot of people say­ing ‘Oh, I made an agent’, and they’re exper­i­ment­ing, and that’s fine, but they’re given these tools without a lot of guid­ance, and the guid­ance that’s out there is of highly vari­able qual­ity.”

Even­tu­ally, Pater­son says, cases will end up in court where legal pre­ced­ents will be set, and the developers will have a duty to mon­itor incid­ents, evolve and improve their pro­to­cols in response. ...

See the full story here; https://www.pressreader.com/australia/the-guardian-australia/20260814/282316801863955

Comments (0) Trackbacks (0)

Sorry, the comment form is closed at this time.

Trackbacks are disabled.