What AI fixes in mental health support ?

Check Your Wellness Program ROI
How much can a Wellness Program save you ?
Eligible Employees (slide to adjust)
500 employees
Average Salary (slide to adjust)
$60,000
$ XXX,679
Your Savings Potential with a Wellness Program
Find My ROI Breakdown

Table of Contents

What AI fixes in mental health support

Only 4% of employees use their organization's EAP. That does not mean the other 96% are fine. It means they are going somewhere else, and increasingly that somewhere is an AI chatbot. Samantha Levin, Wellness Program Manager at Wellness360, on AI, employees and mental health support

Where utilization actually sits

Specific cases are hard to point to, because this happens behind the scenes. It shows up anecdotally, among friends and family as much as anywhere else. Recent surveys indicate employees are significantly more comfortable discussing daily workplace stress with an AI chatbot than with traditional human resources benefits.

EAP utilization is low, and it has been low for quite some time, sitting in the lower single digits for a benefit most companies pay for on behalf of every employee, whether they use it or not. Every company is different, and that figure is an average. The number itself is not new. What is new is finally knowing where people are going instead.

What AI removes and what it does not touch

The five known barriers to EAP use are awareness, stigma, confidentiality, complexity, and limited resources. The chatbot removes the complexity piece. No one is digging through a portal or calling a number during business hours. They type a sentence and get pointed somewhere. It helps with stigma too, since it is a judgment free space before any conversation with an individual. But complexity is the real one it solves, whether that is after hours or a quick question on a lunch break.

What it does not touch is awareness. If someone does not know the benefit exists, a chatbot sitting there quietly is not going to change that. That part sits with the employer: the marketing, the manager conversations, all of it. Making individuals in the company more aware of the resources and benefits they have, and making those accessible, rather than leaving them with a chatbot that offers solutions without access to the resources they most likely need.

AI usage goes wrong when companies and individuals rely on automated tools for high stakes decisions, customer service, or public facing content without human oversight. Without someone constantly reviewing for accuracy and checking that the data matches, it can be harmful to an organization.

The misconception that comes up constantly is people treating every AI conversation as though it is covered by the same confidentiality as a therapist. It is not. On a personal account, that data belongs to the AI company, not even the employer. On an employer provided tool, accessed through a work email, there can be reporting built in that people do not realize is there. The advice is to ask directly whether the employer can see anything before saying something that would not be worth repeating. Shared conversation links from Claude or ChatGPT can also be publicly accessible once shared.

Where utilization actually sits

Many organizations have taken to advising HR to discourage employees from using consumer AI for mental health support. Samantha's position is that individuals should not be turning to AI for mental health support, beyond quick questions and advice. The full story and full picture of what is going on with an individual is not visible there.

What it really comes down to is the crisis situation. The AI can suggest meditation and journaling, but there are deeper conversations going on that the chatbot is not going to see or recognize. So the answer is yes, agree with HR and discourage employees from using AI for mental health support. It remains useful for common exercises like meditation, for quick questions, and for the initial pre-contemplation phase of seeking help, where AI can guide someone to resources in their area, such as the closest therapist office in New York City or crisis numbers. Not for anything deeper.

Two findings from AI chatbot research appear to point in opposite directions. Therabot reports a 51% average reduction in depression symptoms. Stanford found bots respond appropriately only 60% of the time, against 93% for therapists. A benefits leader should view these findings not as a contradiction but as a blueprint for a hybrid mental health strategy. The data shows AI bots are highly effective at reducing symptoms, but they lack the safety and clinical precision of a human therapist. The response is to encourage more in person or virtual human resources instead of an AI chat.

What this says about program design

What has worked is training managers to recognize when someone mentions leaning on a chatbot and giving them a clear next step, a name or a number, rather than noting it and moving on. The recognition piece is the key. That small addition is what actually moves someone from a chatbot conversation into real care.

What has backfired is employers trying to ban or discourage the use of consumer AI outright. People do not stop. They just stop telling everyone around them, and the one signal that someone was struggling disappears. Policy alone does not move the behavior. Manager training paired with a clear point of escalation does.

Underneath all of it, this behavior tells us that programs were built that require someone to identify themselves as struggling before they can get help. That is the stigma piece, and AI removed that step.

Samantha's closing point is that companies should advise individuals to seek services, but seek them in a way that reduces stigma. Much of the difficulty around seeking mental health support comes back to stigma, which is a conversation on its own. Having accessible resources across the organization matters most when someone is in a state of crisis.

Your Workforce Is Your Greatest Asset

Ensure they remain healthy and happy with
Wellness360.
×