Reading time (short enough for your break): 7 minutes
Artificial intelligence is transforming one industry after another, and mental health is no exception. But there's a difference few people know about: not every AI is fit to handle emotional issues.
As AI tools multiply, promising “instant therapy” and “24/7 emotional support,” it's essential to ask: which technology is actually safe and effective for looking after your mental health?
The AI Boom in Mental Health
The global market for mental health apps is expected to reach $17.5 billion by 2030, according to Grand View Research (2025).
In Brazil, the picture helps explain the interest:
- According to the WHO report Depression and Other Common Mental Disorders (2017), Brazil has the highest prevalence of anxiety disorders in the world: 9.3% of the population, compared with a global average of 3.6%
- According to the Mental Health Panorama (Instituto Cactus and AtlasIntel, with data collected in 2023), only 5% of Brazilians are in psychotherapy, while 16.6% take medication on an ongoing basis
- A May 2025 survey by SESI (Brazil's Industrial Social Service) of more than 2,000 Brazilians found that 78% are interested in using digital health services, but only 20% had actually used one in the past year. And 63% say they've never heard of these services or don't know how they work
That creates opportunities, but also risks.
Generative AI vs. Specialized AI: The Difference That Changes Everything
Here's the key point you need to understand before trusting your mental health to any technology:
GENERATIVE AI (ChatGPT, Gemini, Claude, etc.)
What it is: Systems trained on massive amounts of text from the internet to generate coherent answers on almost any subject.
Characteristics:
- ❌ Not built specifically for mental health
- ❌ Learns from sources that aren't all vetted
- ❌ Safety rules are general-purpose, not built around a clinical method
- ❌ Can give contradictory or dangerous information
- ❌ May miss or mishandle risk situations (suicidal thoughts, acute crisis)
- ❌ Ethical limits weren't designed specifically for mental health care
- ❌ Wasn't designed to keep emotional dependence in check
The real risk: A person in crisis may get inappropriate guidance, have serious symptoms played down, or even receive suggestions that make things worse.
A problematic scenario: A user mentions suicidal thoughts → a generative AI offers generic “self-help tips” → the person doesn't seek urgent professional help → the situation gets worse
SPECIALIZED AI (EITA Mentor and similar tools)
What it is: Systems developed specifically for mental health, with targeted training, a stated method and oversight from specialists.
Characteristics:
- ✓ Trained by mental health professionals
- ✓ Uses evidence-based techniques
- ✓ Has protocols for identifying risk
- ✓ Has clearly programmed ethical limits
- ✓ Encourages you to seek professional help when needed
- ✓ Includes anti-dependence mechanisms
- ✓ Undergoes ongoing review by professionals in the field
A safe approach: The technology works as a complement, not a substitute, offering:
- Emotional support between therapy sessions
- Self-knowledge tools
- Emotional regulation practices
- Referrals to professionals when needed
EITA Mentor: A Case Study in Responsible AI
EITA Mentor, developed by psychologist and neuroscientist Anaclaudia Zani (@anaclaudia.eita), is a Brazilian example of AI specialized in emotional support that respects ethical and scientific standards.
What Sets the EITA Method Apart:
1. A Method Developed by Anaclaudia Zani
- Created by a mental health specialist
- Based on structured questions and behavioral neuroscience
- Ongoing review by professionals in the field
2. Safety Protocols
- Spotting warning signs
- Automatic referral to crisis lines, such as 988 in the US, Samaritans (116 123) in the UK and Ireland, and CVV (188) in Brazil
- A clearly limited scope
3. Anti-Dependence Mechanisms EITA is specifically programmed to:
- Encourage user autonomy
- Encourage real human relationships
- Suggest in-person professional care
- Avoid excessive emotional attachment to the technology
4. Transparency About Its Limits The platform is clear about what it can and can't do:
- ✓ In-the-moment emotional support
- ✓ Self-knowledge tools
- ✓ Emotional regulation practices
- ✗ Diagnosing disorders
- ✗ Prescribing medication
- ✗ Replacing in-person therapy
Recognition in the Press
EITA's method has been featured in outlets such as:
- Exame: Interview with Anaclaudia Zani about creating the emotional support AI (in Portuguese)
- Reports on rewiring the brain with help from AI (in Portuguese)
(See @eita.mentoravirtual and mentoravirtual.com.br for full references)
How AI Can (and Should) Be Used in Mental Health
✅ APPROPRIATE USES:
1. First Contact and Screening AI can do an initial assessment and point people toward the right kind of professional help.
2. Support Between Sessions Practicing techniques learned in therapy, logging emotions, tracking progress.
3. Wider Access Offering basic support to people who don't have access to in-person psychotherapy.
4. Psychoeducation Teaching about mental health, symptoms and evidence-based self-care techniques.
5. Ongoing Monitoring Following emotional patterns over time and spotting trends.
6. Preventive Interventions Offering tools before problems get worse.
❌ INAPPROPRIATE USES:
- Replacing a professional diagnosis
- Prescribing medication
- Serving as the only resource in an acute crisis
- Promising to “cure” mental disorders
- Keeping people away from professional help they need
Red Flags: When an AI Is NOT Safe
Be wary of tools that:
- 🚩 Promise to fully replace human therapy
- 🚩 Aren't clear about the limits of what they do
- 🚩 Offer definitive diagnoses
- 🚩 Don't encourage you to see a professional
- 🚩 Aren't transparent about their method
- 🚩 Were built without input from mental health specialists
- 🚩 Have no protocols for high-risk situations
A Practical Guide: How to Use AI Safely for Your Mental Health
BEFORE YOU START:
- ✓ Look into the company and the people behind the tool
- ✓ Check whether mental health professionals are involved
- ✓ Read up on its method and safety protocols
- ✓ Make sure the tool is a complement, not a substitute
- ✓ Understand its data privacy policy
WHILE YOU USE IT:
- ✓ Think critically about the answers you get
- ✓ Don't share extremely sensitive information
- ✓ If the AI suggests seeking professional help, take it seriously
- ✓ Use it as a tool for self-knowledge, not as an “emotional guru”
- ✓ Keep an eye out for emotional dependence on the technology
WHEN TO STOP AND SEEK HUMAN HELP:
- 🚨 Thoughts of suicide or self-harm
- 🚨 Severe symptoms that interfere with daily life
- 🚨 Problematic substance use
- 🚨 Disabling panic attacks or anxiety
- 🚨 Drastic changes in mood or behavior
- 🚨 Any situation in which the AI recommends seeing a professional
If you are in crisis or thinking about suicide, contact your local emergency number right away. In the US, call or text 988 (Suicide & Crisis Lifeline); in the UK and Ireland, call Samaritans on 116 123; in Brazil, call CVV on 188.
EITA Mentor is a responsible model: specialized AI, built on a method, trained by specialists, with clear protocols and an ethical commitment to its users' well-being.
Because your mental health deserves technology that lives up to how much it matters.
Want to get to know an AI built specifically for emotional support? Visit mentoravirtual.com.br and see how technology can support your emotional well-being safely and responsibly.
Follow @eita.mentoravirtual for content on AI and mental health (in Portuguese).
Sources
- Grand View Research (2025). Mental Health Apps Market To Reach $17.52 Billion By 2030. grandviewresearch.com
- World Health Organization (2017). Depression and other common mental disorders: global health estimates. iris.who.int
- Instituto Cactus and AtlasIntel. Mental Health Panorama in Brazil (in Portuguese). institutocactus.org.br
- Correio Braziliense (2023). Only 5% of Brazilians are in therapy, but 16% take psychoactive drugs (in Portuguese). correiobraziliense.com.br
- SESI / Portal da Indústria (2025). 8 in 10 people are interested in using digital health services in Brazil (in Portuguese). portaldaindustria.com.br
Anaclaudia Zani is a psychologist (CRP 06/128106, Brazil) and neuroscientist, creator of the EITA method and of EITA Mentor.