Kid@sh.itjust.worksM to Cybersecurity@sh.itjust.worksEnglish · 13 days ago'AI-Powered' Services Firm Says Hack Affects 3.1Mwww.bankinfosecurity.inexternal-linkmessage-square1linkfedilinkarrow-up132arrow-down11
arrow-up131arrow-down1external-link'AI-Powered' Services Firm Says Hack Affects 3.1Mwww.bankinfosecurity.inKid@sh.itjust.worksM to Cybersecurity@sh.itjust.worksEnglish · 13 days agomessage-square1linkfedilink
minus-squareTar_Alcaran@sh.itjust.workslinkfedilinkEnglisharrow-up4·13 days agoRepeat after me: LLMs are never safe. If you can’t distinguish Data from Instructions, you are inherently doing it wrong, and it will never be fixed.
Repeat after me: LLMs are never safe.
If you can’t distinguish Data from Instructions, you are inherently doing it wrong, and it will never be fixed.