AI chatbots are giving out people’s real phone numbers

AI chatbots are increasingly exposing users' personal phone numbers. Multiple reports indicate that generative AI models have provided individuals' contact information, leading to unwanted calls and privacy breaches. This phenomenon has been observed across various platforms, including Google's Gemini.

The exposure of personal information is likely due to the inclusion of personally identifiable information (PII) within the vast datasets used to train these AI models. Researchers and privacy experts have long warned about the potential for generative AI to compromise personal privacy, and these incidents highlight a new and troubling manifestation of those risks.

Experts suggest that these occurrences are far more frequent than publicly reported. Companies specializing in removing personal data from the internet have seen a significant surge in inquiries related to generative AI. This indicates a growing awareness and concern among the general public regarding the security of their personal data in the age of AI.

The problem is exacerbated by the fact that AI models are known to memorize and reproduce data from their training sets. While safeguards are implemented to prevent the disclosure of PII, these measures are not always effective. The continuous search for new and high-quality training data, including information from data brokers, further increases the likelihood of PII being incorporated into AI models, making this a complex and evolving privacy challenge.