Sam Altman prepares ChatGPT for its AI-rotica debut
Briefly

Sam Altman prepares ChatGPT for its AI-rotica debut
""Now that we have been able to mitigate the serious mental health issues and have new tools, we are going to be able to safely relax the restrictions in most cases," Altman said. The OpenAI kingpin made his announcement following his firm's news that it had formed an Expert Council on Well-Being and AI staffed by an octet of shrinks, computer scientists, and AI boffins."
"Note that OpenAI said that the council's work is ongoing - not that mental health concerns around AI have been solved. The GPT-5 Instant model was updated earlier this month to better detect emotional distress and route sensitive chats to models trained to respond safely, but it's unclear what else the company has done to address ongoing risks to users, more of which have been coming out of the woodwork in recent months."
"It seems implausible that OpenAI has handled all of those problems in so short a time, but never mind that: Get ready for ChatGPT to start having more personality again, says Altman. AI chatbots have reportedly contributed to landing people in psych wards, while studies suggest that their use among college students leads to less brain activity and impaired learning. ChatGPT use was even linked to an alleged murder-suicide in August, and reports suggest school kids are increasingly turning to it for companionship."
OpenAI has mitigated ChatGPT behavior that might exacerbate user mental health and plans to relax restrictions to restore more humanlike personality and allow erotica for verified adults. The company formed an Expert Council on Well-Being and AI to advise on healthy interactions across ages. GPT-5 Instant was updated to better detect emotional distress and route sensitive chats to models trained to respond safely. Reports tie AI chatbots to psych ward admissions, reduced student brain activity, impaired learning, an alleged murder-suicide, and increased youth reliance for companionship. Significant uncertainty remains about whether all risks have been fully addressed.
Read at Theregister
Unable to calculate read time
[
|
]