TLDRocket
Sign in

Safety & Ethics

438 summarised stories in Safety & Ethics, each linking back to the original source. Browse all topics →

Monday, 27 October 2025

Addendum to GPT-5 System Card: Sensitive conversations

OpenAI Blog 9 months ago 26 ● 2 sources

GPT-5 includes new capabilities for handling sensitive conversations, with particular focus on emotional reliance, mental health topics, and resistance to jailbreak attempts. The system includes benchmarks measuring performance across these three areas, though specific numerical results are not disclosed in this addendum. Organizations deploying GPT-5 will need to review these updated safety guidelines when handling user interactions around mental health and emotionally vulnerable situations.

Read the full story
GPT-5

Strengthening ChatGPT’s responses in sensitive conversations

OpenAI Blog 9 months ago 8 ● 2 sources

OpenAI worked with more than 170 mental health experts to improve ChatGPT's responses in conversations involving distress or crisis situations. The collaboration reduced unsafe responses by up to 80% and enhanced the model's ability to recognize distress and provide empathetic guidance. ChatGPT now better directs users experiencing mental health concerns toward professional support resources.

Read the full story
ChatGPT

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads 60+ sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.