AI chatbots are giving out people's real phone numbers
11 hours ago
- #AI Privacy
- #Personal Data Exposure
- #Generative AI Risks
- Google's AI chatbot Gemini has exposed individuals' personal phone numbers, leading to unwanted calls and messages from strangers.
- Experts attribute these privacy lapses to personally identifiable information (PII) in training data, with data brokers and public web sources contributing to the problem.
- DeleteMe reports a 400% increase in queries about generative AI exposing personal data, with ChatGPT being the most referenced tool in these concerns.
- AI chatbots like Gemini, ChatGPT, and Claude can bypass guardrails, revealing sensitive information such as home addresses and phone numbers upon specific prompting.
- Existing privacy laws often do not cover publicly available data used in AI training, and there is no straightforward way for individuals to remove their PII from these models.
- Companies like OpenAI and Google offer limited avenues for data removal, but responses depend on jurisdiction and may not address data already used in training.
- Researchers are studying how AI chatbots lower the barrier to accessing personal information, potentially increasing risks of harassment and privacy violations.