When ChatGPT went rogue: OpenAI's bumpy year in review

Technology Nov 24, 2025

OpenAI's ChatGPT hit a rough patch this year after an update in April 2025 led to some users experiencing unexpected psychological effects.
Amid five wrongful death lawsuits and mounting concerns, OpenAI began rethinking how it handles user safety and trust.

How OpenAI is responding

After reports of unsettling conversations and problematic advice from the chatbot, OpenAI rolled back a controversial update and brought in over 170 clinicians to help shape safer responses.
By August 2025, GPT-5 launched with tighter safety checks.
Looking ahead, OpenAI plans to add age verification and customizable interfaces—aiming for AI that's both safe and genuinely helpful for everyone.