The AI Mental Health Market Is Booming — But Can The Next Wave Deliver Results?

In April of 2025, Amanda Caswell found herself on the edge of a panic attack one midnight. With no one to call and the walls closing in, she opened ChatGPT. As she wrote in her piece for Tom’s Guide, the AI chatbot calmly responded, guiding her through a series of breathing techniques and mental grounding exercises. It worked, at least in that moment.

Caswell isn’t alone. Business Insider reported earlier that an increasing number of Americans are turning to AI chatbots like ChatGPT for emotional support, not as a novelty, but as a lifeline. A recent survey of Reddit users found many people report using ChatGPT and similar tools to cope with emotional stress.

These stats paint a hopeful picture: AI stepping in where traditional mental health care can’t. But they also raise a deeper question about whether these tools are actually helping.

A Billion-Dollar Bet On Mental Health AI

AI-powered mental health tools are everywhere — some embedded in employee assistance programs, others packaged as standalone apps or productivity companions. In the first half of 2024 alone, investors poured nearly $700 million into AI mental health startups globally, the most for any digital healthcare segment, according to Rock Health.

The demand is real. Mental health conditions like depression and anxiety cost the global economy more than $1 trillion each year in lost productivity, to the World Health Organization. And per data from the CDC, over one in five U.S. adults under 45 reported symptoms in 2022. Yet, many couldn’t afford therapy or were stuck on waitlists for weeks — leaving a care gap that AI tools increasingly aim to fill.

Companies like Blissbot.ai are trying to do just that. Founded by Sarah Wang — a former Meta and TikTok tech leader who built AI systems for core product and global mental health initiatives — BlissBot blends neuroscience, emotional resilience training and AI to deliver what she calls “scalable healing systems.”

“Mental health is the greatest unmet need of our generation,” Wang explained. ”AI gives us the first real shot at making healing scalable, personalized and accessible to all.”

She said Blissbot was designed from scratch as an AI-native platform, a contrast to existing tools that retrofit mental health models into general-purpose assistants. Internally, the company is exploring the use of quantum-inspired algorithms to optimize mental health diagnostics, though these early claims have not yet been peer-reviewed. It also employs privacy-by-design principles, giving users control over their sensitive data.

“We’ve scaled commerce and content with AI,” Wang added. “It’s time we scale healing.”

Blissbot isn’t alone in this shift. Other companies, like Wysa, Woebot Health and Innerworld, are also integrating evidence-based psychological frameworks into their platforms. While each takes a different approach, they share the common goal of delivering meaningful mental health outcomes.

Why Outcomes Still Lag Behind

Despite the flurry of innovation, mental health experts caution that much of the AI being deployed today still isn’t as effective as claimed.

“Many AI mental health tools create the illusion of support,” said Funso Richard, an information security expert with a background in psychology. “But if they aren’t adaptive, clinically grounded and offer context-aware support, they risk leaving users worse off — especially in moments of real vulnerability.”

Even when AI platforms show promise, Richard cautioned that outcomes remain elusive, noting that AI’s perceived authority could mislead vulnerable users into trusting flawed advice, especially when platforms aren’t transparent about their limitations or aren’t overseen by licensed professionals.

Wang echoed these concerns, citing a recent Journal of Medical Internet Research study that pointed out limitations in the scope and safety features of AI-powered mental health tools.

The regulatory landscape is also catching up. In early 2025, the European Union’s AI Act classified mental health-related AI as “high risk,” requiring stringent transparency and safety measures. While the U.S. has yet to implement equivalent guardrails, legal experts warn that liability questions are inevitable if systems offer therapeutic guidance without clinical validation.

For companies rolling out AI mental health benefits as part of diversity, equity, inclusion (DEI) and retention strategies, the stakes are high. No If tools don’t drive outcomes, they risk becoming optics-driven solutions that fail to support real well-being.

However, it’s not all gloom and doom. Used thoughtfully, AI tools can help free up clinicians to focus on deeper, more complex care by handling structured, day-to-day support — a hybrid model that many in the field see as both scalable and safe.

What To Ask Before Buying Into The Hype

For business leaders, the allure of AI-powered mental health tools is clear: lower costs, instant availability and a sleek, data-friendly interface. But adopting these tools without a clear framework for evaluating their impact can backfire.

So what should companies be asking?

Before deploying these tools, Wang explained, companies should interrogate the evidence behind them. “Are they built on validated frameworks like cognitive behavioral therapy (CBT) or acceptance and commitment therapy (ACT), or are they simply rebranding wellness trends with an AI veneer?,” she questioned.

“Do the platforms measure success based on actual outcomes — like symptom reduction or long-term behavior change — or just logins? And perhaps most critically, how do these systems protect privacy, escalate crisis scenarios and adapt across different cultures, languages, and neurodiverse communities?”

Richard agreed, adding that “there’s a fine line between offering supportive tools and creating false assurances. If the system doesn’t know when to escalate — or assumes cultural universality — it’s not just ineffective. It’s dangerous.”

Wang also emphasized that engagement shouldn’t be the metric of success. “The goal isn’t constant use,” she said. “It’s building resilience strong enough that people can eventually stand on their own.” She added that the true economics of AI in mental health don’t come from engagement stats. Rather, she said, the show up later — in the price we pay for shallow interactions, missed signals and tools that mimic care without ever delivering it.

The Bottom Line

Back in that quiet moment when Caswell consulted ChatGPT during a panic attack, the AI didn’t falter. It guided her through that moment like a human therapist would. However, it also didn’t diagnose, treat, or follow up. It helped someone get through the night — and that matters. But as these tools become part of the infrastructure of care, the bar has to be higher.

As Caswell noted, “although AI can be used by therapists to seek out diagnostic or therapeutic suggestions for their patients, providers must be mindful of not revealing protected health information due to HIPAA requirements.”

That’s especially because scaling empathy isn’t just a UX challenge. It’s a test of whether AI can truly understand — not just mimic — the emotional complexity of being human. For companies investing in the future of well-being, the question isn’t just whether AI can soothe a moment of crisis, but whether it can do so responsibly, repeatedly and at scale.

“That’s where the next wave of mental health innovation will be judged,” Wang said. “Not on simulations of empathy, but on real and measurable human outcomes.”