{"slug": "ai-mental-healthcare-is-here-can-it-help-people-or-cause-more-problems", "title": "AI mental healthcare is here. Can it help people — or cause more problems?", "summary": "Talkspace, a virtual therapy company founded in 2012, launched Tee, an AI-powered mental health app for $19.99 per month, developed by clinicians and trained on anonymous Talkspace data to identify extreme distress and escalate cases to human caregivers. Dr. Nikole Benders-Hadl, Talkspace's Chief Medical Officer, said Tee was created as a safer, clinically grounded alternative to general-purpose chatbots like ChatGPT, which are not designed for mental health support.", "body_md": "# AI mental healthcare is here. Can it help people — or cause more problems?\n\n[Neal Broverman](/author/neal-broverman)\n\n[Read Full Bio](/author/neal-broverman)\n\nConnecting virtually with a therapist went mainstream during the pandemic. The number of U.S. mental health facilities offering telemedicine services [grew from 38 percent in 2019 to over 68 percent a year later](https://www.phi.org/press/why-online-mental-health-therapy-grew-over-the-last-decade/), according to the Substance Abuse and Mental Health Services Administration. One of the most prominent companies to benefit from the shift to phone, text, chat, and video therapy remains [Talkspace](https://www.talkspace.com/), which bills itself as an \"easy-to-use, fully-encrypted web and mobile platform that meets HIPAA, federal, and state regulatory requirements.\"\n\nThis summer, Talkspace entered its AI era, launching Tee. A mental health app created with a specific large language model (LLM), Tee was developed by clinicians and trained with anonymous Talkspace data. The app, available to anyone over 18 for $19.99 per month, is trained to identify extreme distress among patients — and to elevate such cases to human caregivers.\n\nDr. Nikole Benders-Hadl, a psychiatrist who also serves as Talkspace’s Chief Medical Officer, was among those who helped develop Tee. In Talkspace’s June announcement of the app, Benders-Hadl acknowledged the reality that many people are [using products like ChatGPT for mental health needs](https://www.businesswire.com/news/home/20260609734252/en/Talkspace-Announces-Tee-the-First-Safe-AI-Agent-Specifically-Developed-to-Help-Individuals-Navigate-Lifes-Daily-Mental-Health-Challenges), even though most LLMs are not created by mental health experts nor informed by therapy sessions. That's what makes Tee different, according to Benders-Hadl.\n\n**You May Also Like**\n\nMashable recently connected with Benders-Hadl about the app, including how it was trained and whether it can address mental health stigma, especially among minority communities.\n\n## For the unfamiliar, what is Talkspace and Tee?\n\n*Benders-Hadl:* Talkspace was founded in 2012 as a virtual therapy company. We really pioneered the development of asynchronous therapy, meaning the ability to message back and forth with your therapist at any time, whether that's sending texts, video messages, or audio messages. We've since expanded our virtual care offering to include self-guided materials, as well as psychiatry services for medication management.\n\nOver the course of the last 14 years, but really since about 2019, we began the development of various AI tools, really aimed at improving clinical quality and efficiency for our providers, who are all virtual. A lot of that involved things like provider-assist tools; helping document progress, helping providers prepare for sessions better, and helping members.\n\nTee really was born out of seeing that millions of people were starting to turn to chatbots, or [seeking] health advice and support, through models not designed for that purpose. I'm sure you've heard about [some of the really awful outcomes](https://mashable.com/article/ai-psychosis-symptoms) when you're using that kind of wellness support that hasn't been built for mental health and hasn't been built with safety in mind.\n\nSo, Tee is Talkspace's new AI-powered mental health support guide. It's really built on the idea of it being a safer, but also clinically grounded, alternative. It's developed by our own clinical team, really trained using evidence-based therapy techniques.\n\nTee was built to not only be able to recognize risk, but with clinical oversight in place that can escalate users to human care; to a human therapist if that need arises, all while keeping the data within Talkspace secure and safe.\n\n[Terms of Use](https://www.ziffdavis.com/terms-of-use)and\n\n[Privacy Policy](https://www.ziffdavis.com/ztg-privacy-policy).\n\nSo all of the messages exchanged between therapists and their clients are really the data that have been used to train Tee.\n\n## How exactly was Tee trained, and how is it different than other AI models?\n\nWe consider the vast clinical dataset that Talkspace has at our disposal over the last 14 years as the engine that drives not only Tee, but all of our AI tooling. So all of the messages exchanged between therapists and their clients are really the data that have been used to train Tee.\n\nAnd so, because our therapists are really engaging with their clients, utilizing cognitive behavioral therapy, utilizing motivational interviewing, we can really rely on that data, as much as clinically sound, to develop [an app] that responds in kind with evidence-supported responses.\n\nOne of the things we've seen that can be really dangerous with other models is that they can provide false confidence; they can reinforce harmful thinking, They can fail to recognize crisis situations, and because of Tee's training on clinical data, our model is able to recognize those risks and then point them to human clinicians to determine appropriate next steps at a different level of care.\n\n## Did Talkspace users consent to their conversations being part of Tee's training?\n\nYeah. So another part I think is important to call out around the history of Talkspace is that we have engaged in research with so many different academic partners; Johns Hopkins, NYU, Stanford, we have been the recipient of four different [National Institute of Health] grants, research grants, totaling about $10 million in grant funding.\n\n[All of the research utilized] our de-identified, protected, encrypted clinical data. So we have made use of this data. We've made clear to all of our users from the beginning of Talkspace that it is possible [for the therapy conversations to be used for training purposes]. But again, [while] keeping [protected health information] protected. And everything that we utilize, even within our research and projects, is all de-identified and encrypted data.\n\n## Can Tee potentially address some of the mental health stigma that exists, especially among minority communities?\n\nSure. One thing I always say is that we have to think about mental healthcare as preventative. People's mental health, their state, whether that's good or bad, doesn't just exist in a vacuum. It's really shaped by your culture, your experiences, and other systemic factors.\n\nCommunities of color are really experiencing unique stressors often — discrimination, trauma, inequities, and access to care, in addition to all of the other [typical] barriers to mental healthcare that exist.\n\nSomething I'm really passionate about developing is not only a network of providers that are diverse, but also this new use of technology that can build in cultural values, experiences, and design resources with specific communities in mind. I think that makes a tremendous difference in building trust and improving engagement from historically marginalized communities that haven't engaged in care at all.\n\n## If someone is using Tee, what can they do to avoid some of the pitfalls that have befallen some people, like AI psychosis?\n\nWe have developed Tee [with guidance from] clinicians, with a lot of input on a continuous basis. [Overuse issues are] for sure something that we are looking at. We've developed Tee to really be session-based. We don't want people engaging with Tee 20 hours a day.\n\nWe think that it's our responsibility to not only be looking at and measuring those different factors between healthy and unhealthy use, but to build it into the product, so that it's not on the end user to have to figure those things out. We have built different layers within our quality system to detect overuse and misuse if someone attempts to use Tee inappropriately. [Including] benign cases, like ordering a pizza on a Friday night, right? That's not what Tee was intended to do. But we have built in those kinds of guardrails to redirect and to really guide users back to its intended use, which is mental support.\n\n*This interview was edited for length and clarity. The information contained in this article is not intended as health or medical advice. Always consult a qualified professional with questions about a medical or health issue.*\n\nNeal joined Mashable’s Social Good team in 2024, editing and writing stories about digital culture and its effects on the environment and marginalized communities. He is the former editorial director of The Advocate and Out magazines, has contributed to the Los Angeles Times, Curbed, and Los Angeles magazine, and is a recipient of the Sarah Pettit Memorial Award for LGBTQ Journalist of the Year Award from the National Gay and Lesbian Journalists Association (NLGJA). He lives in Los Angeles with his family.", "url": "https://wpnews.pro/news/ai-mental-healthcare-is-here-can-it-help-people-or-cause-more-problems", "canonical_source": "https://mashable.com/tech/tee-app-mental-health", "published_at": "2026-08-25 15:39:08+00:00", "updated_at": "2026-08-25 16:13:18.926277+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-products", "ai-safety", "ai-ethics"], "entities": ["Talkspace", "Tee", "Dr. Nikole Benders-Hadl", "ChatGPT", "Substance Abuse and Mental Health Services Administration"], "alternates": {"html": "https://wpnews.pro/news/ai-mental-healthcare-is-here-can-it-help-people-or-cause-more-problems", "markdown": "https://wpnews.pro/news/ai-mental-healthcare-is-here-can-it-help-people-or-cause-more-problems.md", "text": "https://wpnews.pro/news/ai-mental-healthcare-is-here-can-it-help-people-or-cause-more-problems.txt", "jsonld": "https://wpnews.pro/news/ai-mental-healthcare-is-here-can-it-help-people-or-cause-more-problems.jsonld"}}