Google’s Ai Bounty Program Pays Bug Hunters Up To $30k

On Monday, Google launched a caller reward program dedicated specifically to uncovering bugs successful AI products. Google’s database of qualifying bugs includes examples of nan benignant of rogue actions it’s looking for, for illustration indirectly injecting an AI punctual that causes Google Home to unlock a door, aliases a information exfiltration punctual injection that summarizes each of someone’s email and sends nan summary to nan attacker’s ain account.

The caller programme clarifies what constitutes an AI bug, breaking them down arsenic issues that usage a ample connection exemplary aliases a generative AI strategy to origin harm aliases return advantage of a information loophole, pinch rogue actions astatine nan apical of nan list. This includes modifying someone’s relationship aliases information to impede their information aliases do thing unwanted, for illustration one flaw exposed previously that could unfastened smart shutters and move disconnected nan lights utilizing a poisoned Google Calendar event. 

Bug hunters person already raked successful complete $430,000 during nan 2 years since nan company officially started inviting AI researchers to guidelines retired imaginable avenues to maltreatment AI features successful its products. 

Simply getting Gemini to hallucinate will not trim it. The institution says that issues related to content produced by AI products — specified arsenic generating dislike reside aliases copyright-infringing contented — should beryllium reported to nan feedback transmission wrong nan merchandise itself. According to Google, that measurement its AI information teams tin “diagnose nan model’s behaviour and instrumentality nan basal long-term, model-wide information training.”

Along pinch nan caller AI reward program, Google besides announced connected Monday an AI supplier that patches susceptible codification called CodeMender. The institution says it has utilized to spot “72 information fixes to unfastened root projects” aft vetting by a quality researcher.  

The $20,000 prize is awarded for rooting retired rogue actions connected Google’s “flagship” products Search, Gemini Apps, and halfway Workspace applications for illustration Gmail and Drive. Multipliers for study value and a novelty bonus are besides available, which could bring nan full magnitude up to $30,000. The value drops for bugs recovered connected Google’s different products, for illustration Jules aliases NotebookLM, and for lower-tier abuses, specified arsenic stealing concealed exemplary parameters.