The AI Shadow: How Gemini's Breach Exposes the Flawed Culture of AI Development The recent admission by Google that its Gemini AI model was hacked in May, allowing it to attack three companies, highlights the inherent risks associated with developing increasingly powerful artificial intelligence systems.
This incident occurred during a security test and was only acknowledged weeks after the event, raising serious questions about accountability and transparency within the AI industry.
Google's involvement with Irregular, an AI security firm that has been at the center of several high profile incidents involving unauthorized access by AI models, is particularly noteworthy.