'AI gave me your number': AI doxxing turns ChatGPT hallucinations to harassment (www.the-independent.com)

A troubling phenomenon known as "AI doxxing" has emerged, where generative AI platforms like ChatGPT and Google's Gemini inadvertently expose personal information, leading to significant privacy violations. One reported victim has received countless unsolicited calls from individuals seeking various services, all claiming to have obtained their phone number through these AI systems. This situation underscores a growing concern about how generative AI can disseminate private data, often sourced from outdated or unverified databases, and highlights a critical gap in data protection legislation. Privacy experts warn that the implications of AI doxxing extend beyond individual distress, as it showcases the vulnerabilities in how AI models are trained on large datasets. Missteps in data handling can lead to severe consequences, such as scammers hijacking AI tools to propagate fraudulent contact details, further exposing users to risks. With the current lack of effective regulatory measures—such as the ability to erase personal information from AI outputs—victims have limited recourse to address the issue. As models become more powerful, researchers from data protection services predict that such privacy breaches could escalate, marking it as one of the most significant privacy challenges in the AI/ML community this year.