State hackers from four nations exploited Google's Gemini AI for cyberattacks, automating tasks from phishing to malware development..
From prompt injection to deepfake fraud, security researchers say several flaws have no known fix. Here's what to know about them.