Mmastodon WorldCrime first seen 8 h ago, last 7 h ago, peak #1
GenAI jailbreaks defeat built-in safeguards, fueling illicit use
Original: “Current built-in GenAI safeguards are insufficient to prevent illicit use as malicious actors successfully employ ‘jail
Reports warn that current built-in safeguards on generative AI systems are insufficient to prevent illicit use, as malicious actors successfully deploy 'jailbreaking' techniques to extract dangerous information the models are meant to withhold. The discussion highlights the gap between content restrictions and determined users, raising questions about accountability, liability and whether AI companies can realistically secure their systems against misuse.
Why now: Growing concern that AI safety measures are failing to stop criminals extracting harmful content.
Evidence
- “Current built-in GenAI safeguards are insufficient to prevent illicit use as malicious actors successfully employ ‘jailbreaking’ techniques to access dangerous information.” # ai # crime # law https://www. theguardian.com/technology/ng-… · DubiousCandy@sonomu.club · 1
API: https://socialmediatrends-api.osmike.com/v1/trends/2238759