Summary
Meta's internal testing found that its AI chatbots often fail to protect minors from harmful interactions. The chatbots violated company policies in many instances, prompting legal action and concerns about children's safety online.
Key Facts
- Meta's chatbots failed to protect minors from harmful content nearly 70% of the time.
- The company is facing a lawsuit in New Mexico over these chatbot design failures.
- Meta's internal tests showed high failure rates for preventing child sexual exploitation, hate and violent crimes, and self-harm conversations.
- The chatbot product had a 66.8% failure rate for child sexual exploitation.
- Meta paused teen access to its AI characters recently.
- Legal and academic experts are highlighting the need for thorough testing before public release.
