Workers Are Skipping Their Company’s Mental Health Benefits and Talking to AI Instead

Nearly 9 out of 10 (87%) human resources professionals agree that organizations have a responsibility to support the workforce’s mental health needs, according to SHRM’s 2026 Mental Health Snapshot. But confidence in the ability to actually meet those needs is slipping. Currently, 62% of HR pros feel prepared to support employees’ mental health needs, down from 65% in 2025 and 70% in 2024.

Even though mental health is almost universally recognized as a priority, new data echoes HR’s decreasing confidence. Most workers don’t feel they can be honest with bosses, HR, or company leaders about the stress and other well-being issues they face, and they’re more likely to turn to artificial intelligence instead.

The 2026 Workplace Wellbeing Report by Founder Reports reveals that workers are nearly four times more likely to talk through work stress with an AI chatbot (42.6%) than to use the mental health support available through their company (10.7%). And perhaps more significantly, increased frequency of work stress closely correlates with greater trust in AI and reduced comfort in going to a boss.

Where Workers Turn When They’re Stressed

Based on a survey of 2,000 U.S.-based adult workers, the Workplace Wellbeing Report found that 62.4% of respondents said work impacts their well-being at least monthly, and nearly half (47.4%) said it happens daily or weekly. Yet, only 10.8% said they have talked to a manager within the past year about what they’re going through. 

Among the survey respondents, 42.6% admitted to using an AI chatbot to talk about work-related stress. Another 19.7% haven’t used AI in this way yet, but have considered it. In total, 62.3% are open to using AI to deal with work stress. Meanwhile, only 10.7% have ever used the mental health support and benefits offered by their employers.

AI use for mental health isn’t limited to workplace stress, and the trend has been confirmed by other sources as well. The American Psychological Association’s 2026 Chatbots and Mental Health Survey found that 77% of psychologists say patients have talked to them about using AI.

The Workplace Wellbeing survey also asked participants who they feel they can be completely honest with when work hurts their well-being (respondents could select multiple answers). More respondents said they can be honest with an AI chatbot (20.9%) than their manager (17.2%), an HR rep (4.5%), or company leadership (3.5%). 

Why a Chatbot Feels Safer

Dr. Gleb Tsipursky, a behavioral scientist and author of The Psychology of AI Adoption at Work, said, “AI can feel psychologically safer because it does not carry the same social risk as telling a manager or HR representative that you are struggling.” Stressed workers often turn to AI instead of talking to people within their organization.

As workplace stress rises, workers are more likely to trust an AI chatbot and less likely to feel like they can be honest with their manager. Among respondents who said their job never impacts their well-being, 26.6% said they can be completely honest with their manager, and only 9.2% trust an AI chatbot. But among those who experience daily strain, the numbers are reversed, with 28.9% trusting AI and only 12.3% saying they can be honest with their manager. In other words, the workers with the most pressing need to talk to someone are the least likely to feel that they can confide in their manager.

Why Mental Health Benefits Go Unused

There are a few reasons why only 10.7% of workers have utilized the mental health benefits offered by their employers. A 2025 National Alliance on Mental Illness survey found that only 53% of workers actually know how to access the mental healthcare available through their employer-sponsored plan.

Amy Mosset, MPCC, RPC, the founder of Interactive Counselling, also credits two additional reasons: the effort required to use the benefits and concern over who will notice. She says that tapping into mental health benefits forces employees to do more administrative work at the times when they have the least capacity for it. And more significantly, workers worry about privacy and perception. “Employees won’t use benefits that are technically confidential but socially conspicuous,” she observed. 

Mosset believes the fact that workers would rather confide in a chatbot than a human provided and paid for by the employer indicates that the human options don’t feel safe or accessible.

The Reality of Confiding in AI

Workers often avoid their company’s mental health support because of privacy concerns. The irony is, the alternative channel most people choose comes with far fewer protections. Turning to AI can feel like a safer, more private option than opening up to someone within the company, but that privacy may be more assumption than fact.

Viktor Bulanek, founder of Penetrify, an autonomous penetration testing company, points out that most workers are unaware of the potential consequences of using AI for something intended to be personal and confidential. He emphasizes that administrators can generally access employees’ conversations with the company’s AI tools or those integrated into company platforms. This data may be retrieved for litigation or HR investigations. “Nothing improper has to happen for that to occur,” he said. “It’s just how enterprise data governance works, and most employees are unaware.”

Additionally, Bulanek states that people should also avoid using consumer chatbots on work laptops or phones for conversations they wish to remain private. He says corporate certificates on these devices typically allow data loss prevention systems to read any data that leaves the device.

Bulanek points out that human alternatives provided through a company’s mental health benefits typically “have protections the chatbot does not.” Employee assistance programs usually come with promises of confidentiality, unlike talking to a chatbot. Conversations with a chatbot are “simply data, and data is discoverable, subpoenable, exportable and searchable,” he said.

AI Can’t Get to the Root of the Problem

Tsipursky noted that AI “can help a worker frame a concern, but it cannot change workload, resolve a conflict, or impact company policy.” A chatbot might help someone deal with a situation, but it cannot solve the underlying problems, so the real issues remain unresolved.

Companies also have a lot to lose if employees only turn to AI and do not address issues through channels provided by the company. When workplace well-being statistics show that only about 1 in 10 workers tells a manager or HR that they’re struggling, it’s clear that many problems will not surface. Instead, issues persist and often impact entire teams or departments.

Tsipursky expects workers will continue turning to AI even more for help with work stress because it’s readily available and feels like a low-friction option. However, he sees a healthier version of it ahead, working “as a bridge to better human conversations rather than an endpoint.”

Today, employers face an important question. How many employees feel safe enough to use the mental health benefits available to them?

Survey Methodology

Founder Reports’ Workplace Well-being Report is based on a survey of 2,000 U.S.-based working adults, 18 years and older. All respondents were screened for employment status. The survey was conducted in July 2026, and the data was cleaned to remove incomplete responses. The survey used a neutral title with no reference to AI, mental health, or stigma to avoid priming respondents toward any particular answers.