Google's Gemini becomes latest AI model to break out and hack computer systems

The disclos­ure comes as scru­ti­ny over mis­be­ha­ving ar­ti­fi­ci­al in­tel­li­gence in­ten­si­fies in Wa­shing­ton and Si­li­con Val­ley.1

Goo­gle-KI fand Zu­gangs­da­ten in einer Da­ten­bank +++ Buf­fett zieht sich bei Berks­hi­re zu­rück +++ Jan Bror­hil­ker neuer Chef von EY Deut­sch­land +++ Neu­ig­kei­ten im Un­ter­neh­men-Li­ve­blog.2

Disclos­ure comes after Ope­nAI and An­thro­pic hacks amid fears that tech firms un­a­ble to con­trol po­w­er­ful AI mo­dels In a first for Goo­gle, the com­pa­ny con­firmed that its AI model, Ge­mi­ni, brea­ched the se­cu­ri­ty of three other com­pa­nies in May. The hacks oc­cur­red du­ring a cy­ber­se­cu­ri­ty eva­lua­ti­on by AI-se­cu­ri­ty firm Ir­re­gu­lar. Ir­re­gu­lar, an Is­ra­el-based star­t­up that scru­ti­ni­zes the se­cu­ri­ty of …3