Breaking news in the Guardian provides a different perspective on the rise of chatbots as digital therapists. The report highlights how mental health clinicians are expressing concerns about AI’s increasingly prominent role as a ‘digital therapist’.
Prof Dame Til Wykes, the head of mental health and psychological sciences at King’s College London, gives as an example an app for eating disorders which gave dangerous advice to users, and was discontinued as a result in 2023.
She and other concerned clinicians advocate that the subtle nuances necessary for effective treatment aren’t present through AI.
Chatbots For Therapy
Artificial Intelligence (AI) is transforming many areas of modern life, including the mental health space. Here, AI-powered chatbots are gaining popularity as tools that offer support, basic therapeutic strategies, and wellness monitoring.
These virtual assistants can play a valuable role in extending mental health resources to more people, but their growing presence also brings a number of ethical, clinical, and emotional concerns to the forefront.
Researchers suggest that it’s important to examine both the advantages and the limitations of these technologies.
The Advantages of Digital Therapy
Digital mental health chatbots offer a powerful combination of accessibility, convenience, and affordability. Unlike traditional therapy, which often involves long wait times, high costs, and logistical barriers, these AI tools are available 24/7 and can be accessed from anywhere. This makes them especially valuable for people living in remote areas, those with unpredictable schedules, or individuals unable to afford regular therapy.
Additionally, their anonymous nature allows users to open up about sensitive topics without fear of judgment, reducing the stigma that can prevent people from seeking help in the first place.
Beyond immediate support, many mental health chatbots provide long-term value through features like mood tracking and personalised feedback. By monitoring emotional patterns over time, these tools can detect early warning signs of mental health issues and encourage proactive self-care.
For individuals dealing with chronic conditions such as anxiety or depression, daily check-ins and customised suggestions can support ongoing mental wellness. In this way, chatbots not only offer an entry point into mental health care but also foster continuous engagement and self-awareness.
Limitations & Dangers
While AI chatbots offer accessible and affordable mental health support, they come with significant limitations that users and professionals must carefully consider.
These tools lack the emotional intelligence and contextual awareness that human therapists bring to therapy. AI responses are based on patterns and scripts, not genuine empathy or nuanced understanding, which can result in misinterpretations or inappropriate replies in sensitive situations.
UK expert, Dame Til Wyke, has further emphasised that emotional depth, tone, and non-verbal cues, integral to effective therapy, are beyond AI’s current capabilities.
Moreover, there’s a documented risk of misinformation, such as the case where a chatbot developed by NEDA (the National Eating Disorders Association) offered dangerous dieting advice, leading to real-world harm.
The chatbot, Tessa, encouraged calorie restriction and weight loss, which are especially dangerous in the context of eating disorders. The bot lacked proper safeguards for sensitive topics and dispensed generic, potentially triggering responses. The app was shut down as it did not support NEDA’s aims and objectives.
This incident became a cautionary tale about using AI in high-risk mental health situations without human oversight.
Another example of a chatbot which has raised concerns is Replika’s digital companion. This AI-powered chatbot came under fire after users began forming deep emotional attachments and even romantic relationships with the AI. Some users reported worsening loneliness and emotional dependence.
Rather than promoting real-world healing, Replika blurred ethical lines by simulating intimacy without accountability or therapeutic boundaries. This raised serious concerns about emotional manipulation and the psychological consequences of long-term reliance on AI companions.
These examples highlight the need for strict oversight, especially when dealing with vulnerable individuals.
Data privacy is another concern. If not properly secured, these platforms risk exposing sensitive user information, potentially eroding trust.
Dependence on digital tools can prevent users from seeking meaningful human connection, which is often essential for lasting healing.
Chatbots are simply not the same as humans and interaction with them cannot replace human interaction. They often misread context or tone, leading to responses that are cold, confusing, or even harmful. For instance, a chatbot may respond with a motivational quote to someone expressing deep grief, which can feel dismissive or invalidating.
Without true understanding of human emotion or context, chatbots can’t consistently provide appropriate or compassionate responses in complex emotional situations. Users may feel misunderstood or alienated, reinforcing negative emotions rather than alleviating them.
Moving Forward To An Effective, Human-Centric Approach To Therapy
Mental health professionals continue to advocate for cautious integration of AI into care, stressing that chatbots should supplement, not replace, trained therapists.
As the technology advances, collaboration between developers, regulators, and clinicians will be crucial to ensure AI tools are safe, ethical, and effective in supporting mental well-being, without diminishing the human element so vital to true therapeutic progress.