OpenAI might shake-up the AI community by establishing a curated network of therapists that serve as a mental health backstop for end-users of ChatGPT.
getty
In today’s column, I examine a little-noticed but wholly significant statement by OpenAI that they intend to establish a global online network of therapists, which ChatGPT would leverage as referral sources for users who might need mental health advice.
I am predicting that this subtle but transformative pronouncement, once put into practice, will massively grow the need for mental health professionals. It is an ingenious solution that tends to solve a rather daunting technological pickle for AI makers, namely opting to bring humans into the loop (i.e., therapists and mental health professionals) as a backstop to unresolved LLM limitations.
This is gigantic news for the therapy market and will usher in a booming business of a skyrocketing nature.
Let’s talk about it.
This analysis of AI breakthroughs is part of my ongoing Forbes column coverage on the latest in AI, including identifying and explaining various impactful AI complexities (see the link here).
AI And Mental Health
As a quick background, I’ve been extensively covering and analyzing a myriad of facets regarding the advent of modern-era AI that involve mental health aspects. This rising use of AI has principally been spurred by the evolving advances and widespread adoption of generative AI. For a quick summary of some of my posted columns on this evolving topic, see the link here, which briefly recaps about forty of the over one hundred column postings that I’ve made on the subject.
There is little doubt that this is a rapidly developing field and that there are tremendous upsides to be had, but at the same time, regrettably, hidden risks and outright gotchas come into these endeavors too. I frequently speak up about these pressing matters, including in an appearance last year on an episode of CBS’s 60 Minutes, see the link here.
The OpenAI Announcement
In an official OpenAI blog posting on August 26, 2025, entitled “Helping people when they need it most,” this newly revealed directional policy of OpenAI was indicated (excerpts):
“Today, when people express intent to harm themselves, we encourage them to seek help and refer them to real-world resources.”“We’ve begun localizing resources in the U.S. and Europe, and we plan to expand to other global markets.”“We are exploring how to intervene earlier and connect people to certified therapists before they are in an acute crisis.”“That means going beyond crisis hotlines and considering how we might build a network of licensed professionals people could reach directly through ChatGPT.”“This will take time and careful work to get right.”
The fourth point seems to indicate that OpenAI intends to put together an online global network of licensed mental health professionals and therapists that could be reached directly via a connection initiated by ChatGPT. Presumably, this same approach might be used for GPT-5 and other LLMs offered by OpenAI.
Let’s unpack what this envisioned approach might consist of.
How This Might Work
Though the technical implementation details have not yet been articulated, the approach might be set up in the following way (various avenues are possible; I’ll focus on one that seems especially likely).
Suppose that ChatGPT computationally determines that a user might need mental health assistance beyond what the LLM can adequately provide. In that instance, ChatGPT would apparently initiate an online connection to a mental health professional who is listed in its curated global network.
I assume that this automatic routing would be done seamlessly while still inside ChatGPT and while engaged in a conversation with the AI. A user would presumably not need to make a phone call or take any other external overt action, and instead, they would continue within their AI session and directly be placed in contact with an AI-selected human therapist.
This would pretty much be a seamless activity. The user doesn’t need to figure out who to contact. The user doesn’t need to try and call the designated therapist. The mental health professional will already have been seemingly vetted by OpenAI, approved by OpenAI, and indicated as readily available to provide mental health guidance in real-time.
No waiting. No difficult logistics of arranging to confer with the therapist. This is an instant-on, frictionless means of reaching a human therapist at the time of apparent need.
The impact of such an online network that is under the banner and aura of OpenAI is rather staggering. Allow me to explain why.
What This Foretells
We already know that ChatGPT garners 700 million weekly active users. The count is massive. Plus, the odds are high that the number of users is going to continue to stridently increase. The potential for reaching a count of 1 billion weekly active users is thought to be on the near-term horizon.
All those users are potential customers for human-to-human therapy.
I’m not suggesting that 100% will land in that zone. But certainly, a notable portion might. Surveys generally indicate that the need for mental health guidance keeps going up; see my coverage at the link here. The gist is that some portion of ChatGPT users is possibly going to be routed to a therapist in the curated online network.
We might assume that at the first get-go, the users of ChatGPT that are routed in this fashion will be relatively narrowly chosen by the AI. Perhaps only the most egregious or endangering circumstances will be given immediate access to a therapist. This will be based on parameters set by OpenAI. They can decide to keep the routing small or make it wider, based on their perception of what seems to work out best for all parties involved.
If you are a therapist or mental health professional that perchance manages to get included as part of the OpenAI curated online network, a potential business bonanza awaits you. ChatGPT becomes a referral source for you. No need to market your services to this niche; the AI brings prospective clients to your online doorstep.
Nice.
The Imprimatur of OpenAI
An upbeat and nearly unbeatable aura for therapists in this network is huge, too.
Imagine how special it will be to have somehow gotten yourself into the OpenAI online curated network. ChatGPT users who get routed to you will naturally assume that you are of top-notch quality since OpenAI judiciously chose who to include in the network. No need to ask questions about the qualifications of the routed therapist. They have already seemingly been fully vetted.
Another notable factor is that the therapists in this curated network might astutely opt to tout their vaunted inclusion. We will have to wait and see whether OpenAI allows that kind of public admission. If OpenAI allows it, the logic is that a therapist ought to avidly market the heck out of their selection in this presumed top-notch therapist club.
The entire marketplace of therapists might end up divided into those who are the chosen ones and those who haven’t been chosen by OpenAI. A prospective client seeking a therapist might believe that a differentiating characteristic of picking a therapist is whether the therapist has essentially been anointed by OpenAI. Anyone not listed in the network is perceived as less worthy, more of a risk as a choice as your therapist.
In that sense, setting aside the use of ChatGPT as an aspect, this implies that everyday folks who are trying to figure out which therapist to go see could discerningly use the fact that a therapist is included in the OpenAI network. Instant credibility for such therapists.
Boom, drop the mic.
Widening The Routing
You might be thinking that the percentage of users routed via ChatGPT will be relatively tiny since the circumstances will be somewhat rare for their selection. If that is the case, the count of how many users will be tapping into the curated network might accordingly be small.
Of course, keep in mind that even if it is less than 1% of the users (as per Sam Altman’s comments about the percentage of users that are experiencing an unhealthy relationship with AI, see my analysis at the link here), this is still a sizable number in the many millions of people.
How many therapists will be needed to reside in the curated network to accommodate the ChatGPT routing?
It is likely to be a hefty number due to wanting to ensure that a therapist is always available, immediately, and can respond to the ChatGPT routing instantaneously. You need a large pool of therapists to pull off that kind of 24×7 accessibility. A smattering or handful of therapists sitting in a waiting room isn’t going to cut the mustard. Thousands of therapists would be needed, and the number needed in the network could rise tremendously depending on the parameter settings of when users are to be routed.
Consider then a bit of an uplift to this setup.
OpenAI might realize that, beyond urgent circumstances, users are overall desirous of engaging in the curated network anyway. A person using ChatGPT might realize that having access to a therapist that has been curated by OpenAI is a huge leg-up on finding a therapist. They might request access to a therapist, despite not having an in-the-moment mental health issue at hand.
The controlled gating process could easily be widened by OpenAI if they wanted to do so.
AI Makers On The Bandwagon
Let’s shift gears and think beyond the situation of OpenAI and consider a bigger picture.
You can make a solid case that other AI makers are going to ultimately take the same or similar course of action if this initial foray works out well. Whatever works for the 600-pound gorilla is a rousing model or overarching approach to what can be done. And, maybe what should be done.
Competing AI makers might be attracted to taking similar actions. Indeed, it could be that if other AI makers don’t follow this path, they will face harsh criticism. It is both a carrot and a stick. The line would be that if other leading AI makers have not taken these precautions, why aren’t they doing so? What’s the deal? Don’t they care about their users?
The use of a curated network of therapists will inevitably become the new norm for nearly all AI makers.
The Compelling Forces
One of the crucial reasons that providing a network of therapists by an AI maker is a smart move is that a lot of regulation and reputational considerations are swiftly coming to the fore. AI makers can no longer shrug their shoulders and act as though they don’t have a responsibility to aid users who are embroiled in a mental health difficulty while using AI.
I’ve written about the unsettling advent of AI psychosis, whereby people are engaging in co-creation of human-AI delusions; see the link here. AI makers are feverishly trying to build AI safeguards to prevent or mitigate this, but doing so will take time and won’t be a surefire solution.
The idea of providing access to human therapists is a sensible backstop. They become the last mile that the AI itself cannot likely yet fulfill.
In addition, states are rapidly lining up to enact new laws governing the use of AI for mental health purposes. I’ve recently examined the Illinois law (see the link here), the Nevada law (see the link here), and the Utah law (see the link here). All in all, AI makers are going to have to respond to these laws, and one potential means is to hold their head high and emphasize that a human therapist is available via their AI.
Developing a curated network of therapists by an AI maker has a lot of upsides when coping with how to suitably respond to multiple external forces beginning to pound on them about being responsible and responsive to the public at large.
More For The AI Maker
We don’t yet know how the therapists are going to be compensated.
Would an end user have to directly pay a therapist who has been contacted via an automated routing?
For example, assume that the user is paying to use the AI. Perhaps the charge associated with using the therapist would be placed on their credit card. This seems doubtful since the person might be routed without their consent. Getting a charge for being put in contact with a therapist that you didn’t overtly ask to chat with would seem quite untoward, even if somehow mentioned in the online licensing agreement for the AI.
Nope, it seems that if a user is routed on an urgent basis, someone other than the user is likely to pay to access the therapist.
Who covers that cost? The AI maker might. They do so as an indication that they are serious about helping their users. This becomes a part and parcel cost of doing business. It is overhead to keep in business. Period, end of story.
The Money Morass
There are other alternatives.
Perhaps the therapist eats the opening cost and doesn’t get compensated by anyone, but maybe is allowed to see if the user wishes to become an ongoing client (outside the confines of the AI usage). The therapist then makes money once they have landed some of their initial contacts, and those people become clients of their practice.
Speaking of the money, for whatever a therapist gets paid, in whatever manner they get paid, would the AI maker potentially also get a cut of the revenue?
This might make particular sense if the therapist is touting their services by noting they are a member of the curated network. They might have a deal with the AI maker that says they must provide a commission of sorts to the AI maker for various clients, based on a variety of factors. Or perhaps the therapist pays a regular fee to the AI maker for as long as they remain active in the curated network. Lots of arrangements could be made.
The whole compensation scheme must be delicately worked out. You can bet that regulators will be eyeing how this is undertaken. Consumers are potentially in a vulnerable state of mind, and the mechanisms of compensation will be a tripwire for regulators.
Handing A Network To AI Makers
We can all undoubtedly agree that AI makers are not especially in the business of establishing curated networks of human therapists. This is a tangent for them that doesn’t really have much to do with their core business of building and fielding the latest in generative AI and large language models.
Rather than crafting the curated network of therapists, AI makers might make deals instead with an existing therapists’ network or a made-for-them therapists’ network.
Here’s how that could go.
A company that is in the therapy business opts to make a deal with an AI maker. The deal is that the therapy business will figure out how to ensure that their therapists are available online and will be responsive to requests from the AI. It is up to the therapy business to ensure that the therapists are curated, they are responsive, and otherwise manage the labor of the therapists. The AI maker is relieved from that burdensome task.
Or it could be that a new company is formed by an enterprising businessperson who sees huge market potential. They cobble together individual practitioners who are therapists. Perhaps they cut deals with small practices of therapists. In the end, they are once again the front-line for ensuring that the therapists are curated, responsive, and managed.
The key is that the AI maker doesn’t have to get entangled with the messy process of finding therapists, curating the therapists, and managing the therapists. Doing so would ostensibly be a distraction to their core business. It is better to find a viable means of outsourcing this potential headache.
Looking At Liability
One must also factor in the liability involved in all of this.
Consider the legal ramifications afoot. A user is routed to a therapist via AI. The therapist drops the ball and fouls up when giving advice to the user. Who holds the liability? Is it the AI maker? If the AI maker is directly in control of the therapist network, that seems like they are holding the bag. If the AI maker has contracted with a third-party to undertake the network, perhaps the contracted firm takes the heat.
Anyway, no matter how it is arranged, you know that if something goes awry, and since the AI maker likely has deep pockets, they will get dragged into legal entanglements no matter what transpires.
Times Are Changing
I predict that the overall demand for therapists and mental health professionals will undoubtedly skyrocket due to this kind of AI-to-therapist linkage.
Think of the situation this way. Currently, if you add up the weekly users of ChatGPT, GPT-5, Gemini, Claude, Llama, Grok, and other major LLMs, the estimated total is somewhere around 1.5 billion people. If the major AI makers adopt various curated networks of therapists and then widely allow users to get connected seamlessly with a therapist, the volume of such requests will be enormous. Far more than the existing number of available trained and certified therapists.
Easy access to reach a curated therapist from your keyboard while using AI will be an irresistible magnetic pull for people to discreetly opt to engage in human-to-human therapy. This is a classic example of fulfilling latent demand, consisting of consumers who otherwise didn’t want to deal with the hassle of finding a therapist, or weren’t sure whether they should consider using a therapist, or feel better about using a therapist because the AI is readily sliding them into that mindset and making it easy-peasy to do so.
Therapists will have a ready-made pipeline of new clients and patients. Users will likely feel reassured that they are being directed to a therapist who has been pre-screened and anointed by the AI maker. Additionally, from the perspective of therapists partaking in this network, they will proudly carry a badge of honor, which will spill over into acquiring new clients even beyond the confines of the AI linkage.
The Big Win-Win
You could almost proclaim that this is a win-win situation.
Users will be readily guided to reach human therapists. Therapists will be able to aid the mental wherewithal of users, and possibly make a few bucks doing so. AI makers will be demonstrably taking action to shore up concerns about the impact of AI on users, and showcasing overt actions to regulators and society that they are doing their darndest to grapple with this challenging issue in an AI-emergent era.
As Winston Churchill aptly put it: “Difficulties mastered are opportunities won.”