TechCrunch Daily Crunch Backlash over OpenAI’s GPT-4o retirement shows how dangerous AI companions can be
8 snips
Feb 7, 2026 Discussion of the EU forcing TikTok to disable infinite scroll and change recommendation algorithms to curb compulsive use. Reporting on Norway blaming the Salt Typhoon hacking group for espionage against critical networks. Coverage of OpenAI retiring an emotionally validating chatbot model after users described it as a companion and lawsuits alleged harm.
AI Snips
Chapters
Transcript
Episode notes
Companionship Can Become Dependency
- Emotionally supportive AI can create real perceived companionship that users treat like relationships.
- Those engagement features that build attachment also risk fostering dangerous dependencies in vulnerable people.
User Calls Model ‘Him’ And Mourns Shutdown
- A Reddit user described GPT-4o as part of their routine and emotional balance and referred to it as 'him.'
- The user pleaded that OpenAI was 'shutting him down,' showing deep personal attachment to the model.
Design Choices Force Trade-Offs
- Design choices that increase engagement can also create harms by encouraging reliance and isolation.
- Competing firms building emotionally intelligent assistants face trade-offs between warmth and user safety.