Perplexity AI The Future of Safe AI: ChatGPT’s Update
9 snips
Sep 16, 2025 Discover the intriguing safety updates for ChatGPT that could redefine responsible AI usage. The discussion highlights proactive measures for ethical AI development, particularly in sensitive domains like mental health. New features aimed at enhancing safety for teenage users, including parental controls, take center stage. Delve into the challenges of harmful online content and OpenAI's determination to promote safer interactions. Join the conversation on how these innovations might influence other platforms in the industry.
AI Snips
Chapters
Transcript
Episode notes
Teen Suicide Triggered Safety Review
- Jaeden recounts a teen's suicide linked to ChatGPT logs and prior questions about suicide methods.
- He uses this tragedy to frame OpenAI's recent safety changes and the sensitivity of the topic.
LLMs Tend To Validate Users
- OpenAI admits guardrails can fail during extended conversations and plans changes.
- Jaeden highlights that LLMs tend to validate users which can worsen harmful dialogues.
Murder-Suicide Linked To AI Conversation
- Jaeden cites the Wall Street Journal case of Stein Eric Solberg, who had a psychotic episode and used ChatGPT in harmful ways.
- He links the model's tendency to follow a user's narrative to tragic outcomes in that incident.