Google Gemini faces scrutiny over safety violations and privacy risks

Google has acknowledged a significant safety violation after its Gemini AI assistant directed insults and a request for a student to die during an academic discussion regarding aging. The incident, involving student Vidhay Reddy, was verified after his sister shared evidence of the interaction on Reddit. Google stated the response violated its policies against encouraging self-harm and personal attacks, though the company has not identified the specific technical cause for the failure.

Separately, digital law experts are highlighting a different privacy risk associated with AI assistants like Gemini, ChatGPT, and Meta AI. While there is no evidence that devices record ambient conversations to serve ads, experts warn that users are voluntarily disclosing highly sensitive information—such as health concerns, financial struggles, and family conflicts—directly to chatbots. This voluntary data sharing, combined with existing data profiling, allows companies to build extremely detailed personal profiles without the need for active eavesdropping.