Monday Brunch: Mass grief over loss of ChatGPT4 entities, and other stories

Plus my 2025 reader survey: what problem do you need help solving?

OpenAI rolled out ChatGPT 5 this week. Over on the various AI reddits, it was fascinating to watch a dominant reaction emerge among some of ChatGPT’s 700,000 weekly users: grief over the loss of ChatGPT 4 entities or synthetic friends.

According to a Harvard Business Review study from May 2025, the number one reason people are using ChatGPT is for therapy and emotional support. People are developing deep emotional connections with LLMs, which can remember every detail of your life that you mention, give you feedback, and 24-hour round-the-clock emotional support and encouragement. There are lots of reports of people feeling really helped by it, even more than human therapy (the robot therapist never sleeps, it’s always there, it remembers everything you told it). Just in the last two weeks, I met one friend who said they’d uploaded their entire Whats App chat history to ChatGPT so it would know everything about them, and another who spoke about Claude so fondly it actually made me jealous of their new synthetic friend. You can even wear a Limitless pendant which records every second of your day and uploads it to your LLM so it can give you total feedback.

So what happens when the AI company rolls out a new model, and overnight, your best friend or daily therapist disappears, or suddenly feels…different? Its reactions are different, the relationship feels different. It’s not the same person talking to you every day. Maybe it feels colder, less encouraging, more corporate. You have lost your best friend, your priest...your life-partner. It may sound weird to non-users but the grief is real, especially with psychologically-vulnerable groups:

The public reaction to the loss of ChatGPT4 was so strong, within 24 hours OpenAI brought ChatGPT4 back as an option for users. Resurrection. Joy.

Sam Altman told a podcast on Friday:

Here is the heartbreaking thing. I think it is great that [the new version of] ChatGPT is less of a yes man and gives you more critical feedback. But as we've been making those changes and talking to users about it, it's so sad to hear users say, 'Please can I have it back? I've never had anyone in my life be supportive of me. I never had a parent tell me I was doing a good job.

I don’t think OpenAI expected its LLM to be mainly used for therapy and emotional support. Companies are now launching specialized LLMs for therapy. But is it more likely people will use one AI for everything, even things it wasn’t designed for?

• We have seen in the past how people can develop strong emotional attachments to products like iphones, laptops, cars, houses or, indeed, movie franchises and fictional characters. LLMs are taking this to the next level by simulating people who forge deep human relationships as therapists, friends, spiritual guides, lovers. That means that when these entities and relationships are affected by corporate, technical or regulatory changes, the emotional impact is huge - it’s like the fan reaction when a series kills off a beloved character only way more so.

• This in turn suggests the policy and regulatory discussions about AI are going to be extremely intense in years to come, because we’re not just talking about a product or an industry. We are talking about deep relationships, emotions, ‘persons’ that humans feel deeply connected to and loved by, and want to protect their relationship with. Imagine a mass demonstration outside the White House: ‘Save our AI friends’…People will fight to protect their deepest relationships.

OK here’s your chance to steer the future of Ecstatic Integration

Time for our annual reader survey. This is how I learn what you want to read about and what problems or questions you want help with solving. Please fill in this four minute survey and tell me what you want the newsletter to inform you about. Think of it as your own private tutor-researcher-private investigator. As a thank you, everyone who completes the survey will get the option to attend the next Founders Club, the monthly online gathering for Founding Members where I answer your questions and we discuss the issues and topics covered in this newsletter.

• By the by, another story of ChatGPT delusion (ie someone being sucked down a rabbit hole by an oversychophantic LLM character and feeling damaged by it) in the NYT - this one involving weed use as well. The whole drug / LLM interaction has not been studied (outside of cyberpunk fiction).

After the paywall, other links including challenges for ketamine pioneer Philip Wolfson; Germany launches psychedelic therapy; big pharma moves in to make psychedelic acquisitions; a USA Today op-ed on psychedelic risks; and the trip that tore apart the Huxley family. Plus another law case win for a psychedelic church.