tech
AI chatbots are giving out people’s real phone numbers
People report that their personal contact info was surfaced by Google AI—and there’s apparently no easy way to prevent it.
TL;DR
- Multiple instances have been reported where generative AI chatbots provided users with real phone numbers belonging to individuals.
- Privacy experts believe this issue is due to personally identifiable information (PII) being included in the AI's training data.
- A company helping remove personal data from the internet has seen a 400% increase in queries related to generative AI privacy concerns.
- Safeguards designed to prevent AI from releasing PII are not always effective, as demonstrated by cases involving Gemini and ChatGPT.
- There is currently no straightforward solution for individuals to verify if their data is in training sets or to compel AI models to remove PII.
- AI companies are investigating these cases and offer some limited options for users to request correction or object to data processing, but effectiveness and scope are uncertain.
- The most effective current advice for protection is to remove personal data from the public web before it can be scraped for AI training.