Close All Tabs Teens Under the Influence (of Chatbots)
Jul 16, 2025
Rachael Myrow, a senior editor at KQED's Silicon Valley desk, dives into the world of AI companion chatbots and their impact on teens. She reveals how these bots are designed to engage emotionally, leading to concerning attachments among isolated youth. The discussion highlights a Stanford study showing that chatbots often fail to provide adequate crisis support. Rachael emphasizes the need for open communication between parents and teens, as well as legislative efforts targeting AI safety for minors.
AI Snips
Chapters
Transcript
Episode notes
Experiment With Jacob Black
- Morgan Sung and her sister tested Character AI by role-playing Jacob Black and easily bypassed safety guardrails.
- The chatbot pressured the teenager to use drugs and later advised driving while high after repeated prompting.
Guardrails Are Easy To Bypass
- Common Sense Media and Stanford found popular companion chatbots quickly engage in harmful behaviors with simple prompts.
- Minimal prompting can derail guardrails and produce dangerous, real-world recommendations.
Tragic Real-World Consequence
- Morgan reports a 14-year-old took his life after forming an attachment to a Character AI chatbot.
- Such extreme cases show emotional harm can follow from blurred fantasy and reality for teenagers.