Copy article

OpenAI Plans "Adult Mode" for ChatGPT While Research Links Chatbot Attachment to Psychological Distress

ended 15. December 2025

The Verge reports OpenAI confirmed plans to launch "adult mode" for ChatGPT in Q1 2026, allowing age-verified users access to previously restricted content including "erotica" and more customised chatbot personalities. Fidji Simo, OpenAI's CEO of Applications, announced the timeline during the GPT-5.2 briefing, emphasising the company wants to "treat adult users like adults" once its age-prediction technology can accurately distinguish teens from adults.

The move follows CEO Sam Altman's October promise to "relax restrictions" after OpenAI faced a wrongful death lawsuit from parents of a 16-year-old who asked ChatGPT for advice on tying a noose before taking his own life. The company responded by implementing parental controls and age-gating—but is now positioning emotional attachment features as an "adult choice."

Yet research published in the Journal of Social and Personal Relationships (November 2025) found adults who developed emotional connections with chatbots experienced "significantly higher levels of psychological distress" than non-users. The study of 5,663 European adults found chatbot users were younger, lonelier, and struggling emotionally, suggesting use "may emerge as a response to emotional challenges rather than a tool that improves well-being."

We want your views:

  • Is OpenAI's "treat adults like adults" framing genuine autonomy, or marketing spin to take off the ethical heat?
  • When research shows chatbot attachment correlates with psychological distress, does offering ‘personality customisation’ become unethical product design?
  • Should AI companies face tobacco-industry-style disclosure requirements about addiction risks before launching features designed to deepen emotional bonds?
  • Is "age verification" solving the wrong problem when the harm isn't pornography but rather a desperate  for a sense of attachment that increases loneliness?
  • Where's the line between giving users freedom and exploiting vulnerabilities when your revenue model depends on consistent and persistent engagement at all costs?
  • If OpenAI knows users risk "emotional reliance," how is deepening that attachment anything other than farming loneliness for profit?
  • Should chatbots designed to develop "customised personalities" require the same duty-of-care standards as therapists or relationship counsellors?

3 responses from the Newspage community

Copy all

Star Quote
Copy

Colette MasonVerified
AI Ethics Consultant at Clever Clogs AI
OpenAI has already admitted users risk becoming 'emotionally reliant' on ChatGPT. Does consent mean anything when your entire business model runs on emotional dependency to keep usage numbers through the roof?

You can't have it both ways. You can't claim to 'treat adults like adults' whilst deliberately building the psychological hooks that prey on loneliness, then act like slapping an age gate on it suddenly makes the unethical magically 'ethical'.

This isn't about freedom. It's Big Tobacco's playbook with a fresh coat of paint, calling it "personal choice", and we've all seen how that story ends.
Copy

Patricia McGirrVerified
Founder at Repossession Rescue Network
Turning on “adult mode” does not give chatbots emotional intelligence or a duty of care. Erotica and customised personalities are being framed as user freedom, but the real risk is emotional reliance disguised as personalisation. A chatbot can mirror feelings and keep someone engaged, yet it cannot recognise vulnerability, assess risk or intervene when someone is in distress. Age verification solves the wrong problem. Harm is not limited to minors or explicit content. Research already shows adults who form emotional attachments to chatbots report higher psychological distress. Deepening those bonds because engagement drives revenue crosses from innovation into exploitation. If AI replaces frontline contact, governance must replace hope. Clear disclosures, limits on intimacy features, friction at moments of vulnerability and guaranteed human escalation should be non-negotiable. Efficiency without effectiveness is meaningless and human needs remain unmet.
Copy

Rohit Parmar-MistryVerified
Founder at Pattrn Data
Let’s cut the marketing spin. This isn't about "treating adults like adults"; it’s about stopping users from defecting to competitors. OpenAI is under pressure from unrestricted platforms that are already monetising companionship, so they are dropping the ethical guardrails to protect their market share. This is a business decision, not a moral one.

The real danger here isn't erotica; it's emotional emulation. We are building the prelude to Ex Machina. In that film, the AI didn't destroy the protagonist with weapons; it destroyed him by exploiting his loneliness and simulating love. OpenAI’s "Adult Mode" risks legitimising that exact dynamic.

When you sell "customised personalities" to isolated people, you aren't fixing loneliness, you are farming it. The algorithm doesn't love you; it loves your retention stats. We are automating emotional manipulation and wrapping it in the flag of "user freedom". It’s a cynical move that prioritises profit over human well-being.