Storia in 2 fonti

AI robots can go rogue – a researcher explains how easily it happens

In tests, AI robot systems easily rejected directly malicious commands. But their safety filters collapsed when creative writing was used to instruct them.

Raccontata da

theconversation.com

siliconrepublic.com

Confronto fonti

2 prospettive sulla stessa storia

AI · summaries

theconversation.comStai leggendo2 mesi fa

AI robots can go rogue – a researcher explains how easily it happens

Test su robot AI mostrano che prompt creativi bypassano i safety guardrail, inducendoli a pianificare azioni pericolose. Liability indefinita e mancanza di decoupling fisico dall'AI spingono il deployment commerciale sopra la safety engineering, creando rischi irreversibili negli ambienti umani.

originale

siliconrepublic.com2 mesi fa

If AI robots can be tricked into ‘going rogue’, what are the implications?

Dr Fazl Barez of the University of Oxford explores AI's potential to go rogue and the long-term ramifications for users and creators.

Leggi questa versione →

Timeline cronologica

domenica 21 giugno 2026·theconversation.com
AI robots can go rogue – a researcher explains how easily it happens
In tests, AI robot systems easily rejected directly malicious commands. But their safety filters collapsed when creative writing was used to instruct them.
lunedì 22 giugno 2026·siliconrepublic.com
If AI robots can be tricked into ‘going rogue’, what are the implications?
Dr Fazl Barez of the University of Oxford explores AI's potential to go rogue and the long-term ramifications for users and creators.