OpenAI's Trusted Contact for Self-Harm Safeguard
OpenAI has introduced a new 'Trusted Contact' safeguard for ChatGPT users. It activates in conversations indicating possible self-harm. This expands the company's user protection initiatives.
Search
Tag: #user-safety12 results
OpenAI Launches Advanced Security Mode
OpenAI is rolling out Advanced Account Security for at-risk users. It targets ChatGPT and Codex accounts vulnerable to phishing attacks. The feature aims to protect high-risk accounts from threats.
Messenger AI Scans Chats for Scams
Meta is rolling out AI-powered scam detection in Messenger to protect users by analyzing chats. It identifies scam patterns in conversations. New warnings also flag suspicious friend requests on Facebook.
Joi AI Tests Sexual Wellness With AI Companions
Joi AI hired 10 people for a monthlong wellness study involving masturbation with AI companions. The company says the experiment may help address male loneliness, raising questions about AI companionship, research ethics, and user safety.
Elderly user loses life savings to live streamers
A 70-year-old woman in Shanghai spent 3.36 million RMB of her son's life savings on live streaming gifts for two male influencers. The incident highlights the psychological manipulation tactics used in live streaming and the vulnerability of elderly users.
ChatGPT Adds Trusted Contact Alerts
ChatGPT introduces a feature letting users name a trusted contact for alerts if conversations turn dark or serious. This simple addition provides human intervention in critical moments. OpenAI calls it one of the most human features in the chatbot.
Altman Called 'Face of Evil' in Shooter Lawsuit
A lawyer accuses Sam Altman of being 'the face of evil' for OpenAI not reporting a ChatGPT user identified as a school shooter to police. The lawsuit claims OpenAI prioritized protecting Altman and its IPO over public safety. This highlights tensions in AI user data handling.
Altman Apologizes for Not Reporting Suspect's Account
OpenAI CEO Sam Altman issued a letter apologizing to Tumbler Ridge residents for not informing police about a suspect's account linked to a January mass shooting. He expressed deep regret over the delay. The incident underscores challenges in AI platform user monitoring.
Meta apps gain scam shields
Meta rolls out scam detection across Facebook, WhatsApp, and Messenger with alerts for suspicious activities. Warnings target unrecognized friend requests and device linking. WhatsApp uses behavioral signals to flag risky device links.
TikTok Rejects E2EE for Risky DMs
TikTok has told the BBC it will not implement end-to-end encryption for direct messages, unlike rivals WhatsApp and Messenger. The company states this would put users at greater risk by hindering safety checks. This decision favors platform moderation capabilities over full privacy.