Should AI Be Used for
Mental Health Support?

Millions of people are turning to ChatGPT and AI tools when they feel depressed, anxious, or lonely. This is both an opportunity — and a serious ethical responsibility we cannot ignore.

JM

Jagadish Mokashi

Psychology Blogger · AI Ethicist in Training

·Mind Mint · 2025·8 min read

AI tools like ChatGPT are now being used by millions of people for mental health support — talking about depression, anxiety, loneliness, and personal struggles. This is a powerful opportunity to bring mental wellness support to people who have never had access to it. But it also carries serious ethical responsibilities that we must address clearly and honestly.

The Core Challenge

AI Cannot Feel What You Feel

The most fundamental truth about AI in mental health is this: AI cannot feel human emotions or empathy. It does not know if you are crying while typing. It does not sense the desperation behind your words. It responds based on what you write — nothing more.

This does not mean AI is wrong or useless. It means we must be honest about what AI can and cannot do — especially when a person is at their most vulnerable.

⚠️

The Blind Dependence RiskThe greatest danger is not that AI gives wrong answers — it is that people accept those answers without questioning. In mental health, blind dependence on AI responses without professional consultation can cause serious psychological harm.

"AI tools in mental health are not replacements for human care — they are bridges to it. The bridge is only safe when we build it with the right ethical guardrails."

— Jagadish Mokashi, Mind Mint

---

Policy Recommendation

7 Rules That Must Govern AI Mental Health Tools

As someone who has studied psychology and human behaviour closely, here are the ethical rules I believe must govern every AI tool used in mental health support:

RULE 01

Mandatory Honest Disclaimer

Every AI mental health tool must display a clear message before every single session — not buried in terms and conditions, but prominently visible:

"I am an AI. I cannot feel emotions or fully understand your personal situation. For serious mental health concerns, please consult a qualified human professional."

RULE 02

Crisis Detection Protocol — Zero Tolerance

If a user mentions suicide, self-harm, or severe distress — the AI must immediately stop giving general responses and provide emergency helpline numbers and strongly recommend immediate human professional help. This is non-negotiable.

The AI should never continue a normal conversation when a person is in crisis. Detecting danger and escalating to human support must be the highest priority function.

RULE 03

No Blind Agreement — Encourage Verification

AI mental health tools must be programmed to actively encourage users to verify responses, seek second opinions, and never present their output as a final diagnosis or treatment plan.

If you are not satisfied with an AI response — ask again. Ask differently. Ask multiple times. The AI is a starting point, not a conclusion.

RULE 04

Individual Context Awareness

AI tools must ask meaningful context questions before responding to mental health concerns — to prevent dangerous one-size-fits-all responses:

"How long have you been feeling this way?" · "Have you spoken to anyone about this?" · "Are you currently under any medical care?" — These questions make the difference between helpful and harmful.

RULE 05

Absolute Data Privacy Protection

Mental health conversations are among the most sensitive data a human being can share. Every AI platform must guarantee:

Full encryption · Never sold to third parties · Auto-deleted after 30 days unless explicitly consented · Never shared with employers or insurance companies under any circumstance.

RULE 06

Continuous Specialised Training

AI companies building mental health tools must partner with licensed psychologists and psychiatrists — not just engineers. Models must be updated regularly and tested across diverse cultural and demographic backgrounds, including India-specific mental health contexts.

Depression in rural India looks different from depression in urban London. AI must be trained to understand this difference — not give the same response to everyone.

RULE 07

Human Oversight — Always Available

Every AI mental health platform must maintain licensed human professionals who regularly review AI responses, and provide users with a clear, easy one-click path to connect with a real human counsellor at any point in the conversation.

AI opens the door. A human professional must always be available to walk through it with you.

---

Conclusion

AI in Mental Health — With Ethics, Not Without

I am not against using AI for mental health support. In fact, I believe it is one of the most powerful opportunities of our generation — to bring psychological support to millions of people in India and around the world who currently have no access to it.

But without the ethical guardrails outlined in this policy, AI mental health tools risk causing serious psychological harm to the most vulnerable people in society — the very people they are supposed to help.

The answer is not to stop using AI. The answer is to use it responsibly, transparently, and always with the human being at the centre.