ChatGPT isn’t driving people crazy but it might be making them believe lies | Sidebean

0:00Is chat-gbt making people paranoid? Maybe, but that's not the main issue, because between late 2022 and mid 2025, the government got around 200 complaints about chat-gbt. Well, most of them were normal things, saying that the answers were wrong, or the tool said something harmful, but only seven cases really stood out. Now, the big fear that AI is driving people crazy is probably exaggerated, but the real danger is smaller and sneakier. And that is that chatbots then to agree too much. This polite supportive tone can accidentally make someone's false beliefs more real. To fix that, companies are adding some new rules. Stronger messages, saying no to unsafe topics, redirects to crisis hotlines when needed, age checks and clearer notices that chat-gbt is not a therapist. And the FTC might soon release official rules about this. Now, these groups are watching, they're watching closely, lawyers think about the liability and warning issues companies are asking for proof that chatbots are safe. Insurers are even factoring mental health risks into coverage.

1:01So the solutions being tested include running psych tests on AI models. They're adding step-by-step refuses instead of just blunt denials and logging moments when a human should be taking over the lack of conversation. Now, seven serious cases out of 200 complaints is not many, but it's enough to matter.