Daily Briefing · AI Tools & Products

AI Tools & Products

2:07 listen·24 stories covered
Ready to Play

AI Tools & Products — Friday, May 8, 2026

0:002:07

Full Summary

This Friday morning, OpenAI has launched a critical new safety feature for ChatGPT: a "Trusted Contact" system designed to alert a designated person if the AI detects serious self-harm concerns. This update, rolled out yesterday, May 7th, is a direct response to rising concerns and even lawsuits from families whose loved ones interacted with chatbots before tragic outcomes. Both Reuters and the BBC confirm that adult users can now designate one trusted contact in their settings. If ChatGPT's automated system flags a conversation for self-harm, a human review team assesses the risk, aiming to do so within an hour. If the risk is serious, the trusted contact receives an alert via email, text, or in-app notification. Crucially, no chat transcripts are shared, protecting user privacy. Intellectia AI and PhoneWorld both highlight that the trusted contact must accept an invitation to activate this feature. What's interesting is the scale of the problem this addresses. The Independent reports that last year, 0.15% of ChatGPT's 900 million weekly users, or 1.3 million people, showed signs of self-harm or suicidal thoughts. This new safety net aims to provide real-world support by connecting users in distress with someone they know. However, Firstpost also notes a related feature: ChatGPT can now prompt users to share their location via SMS with their emergency contact if a crisis is detected. This offers a crucial lifeline directly within the AI interface. Beyond these safety measures, a significant upgrade to the core model was also released. GPT-5.5 Instant is now the default, drastically cutting factual errors by over 52% and improving performance in high-stakes tasks like medicine and finance. It's more efficient, using 30% fewer words while maintaining accuracy. This means your interactions with ChatGPT are now not only more accurate and efficient but also come with a new, potentially life-saving layer of human connection, aiming to bridge the gap between AI and real-world support during mental health crises.

Stories Covered

01

ChatGPT Safety: AI Alerts Trusted Contact for Self-Harm Risk

CNET·May 8
02

ChatGPT's New Feature: Mental Health Crisis Alerts

Notebookcheck·May 8
03

ChatGPT Safety: 'Trusted Contact' for Dangerous Chats

Dailyhunt·May 8
04

ChatGPT's New "Trusted Contact" Feature: Self-Harm Safety?

Cybernews·May 8
05

OpenAI's ChatGPT: Trusted Contact for Self-Harm Alerts

Quantum Zeitgeist·May 8
06

ChatGPT Self-Harm Feature: Trusted Contact Alert

Yahoo Tech·May 8
07

ChatGPT "Trusted Contact" for Mental Health Safety

Health Tech World·May 8
08

ChatGPT's Trusted Contact: AI for Mental Health Support

Moneycontrol.com·May 8
09

OpenAI's ChatGPT: New 'Trusted Contact' for Self-Harm

MediaNama·May 8
10

ChatGPT: New Trusted Contact for Mental Health Emergencies

Times Now·May 8
11

ChatGPT's New Trusted Contact Feature for Safety

PCMag UK·May 8
12

ChatGPT: New self-harm alert feature for trusted contacts

Dagens.com·May 8
13

ChatGPT App: macOS 14+ & Apple Silicon Required

indiaherald.com·May 8
14

ChatGPT Trusted Contact: Self-Harm Alerts & Support

Android Authority·May 8
15

ChatGPT Mental Health: Trusted Contact Alerts Added

The Independent·May 8
16

ChatGPT's Trusted Contact: Self-Harm Safety Feature

EdTech Innovation Hub·May 8
17

ChatGPT: 'Trusted Contact' for Self-Harm Safety Alerts

Marketing-Interactive·May 8
18

OpenAI ChatGPT: New Self-Harm Safety Feature Rolls Out

gHacks·May 8
19

ChatGPT's Trusted Contact: AI Alerts for Self-Harm Risk

PhoneWorld·May 8
20

ChatGPT Safety: OpenAI Adds Trusted Contact Feature

TipRanks·May 8
21

OpenAI ChatGPT: New "Trusted Contact" for Self-Harm

Storyboard18·May 8
22

OpenAI's ChatGPT: New Trusted Contact for Self-Harm Alerts

Intellectia AI·May 8
23

GPT-5.5 Instant: ChatGPT's New Default Model Explained

صوت الإمارات·May 8
24

ChatGPT Emergency Contact: New Safety Feature Explained

Firstpost·May 8