When AI Is the Wrong Tool
Confident AI users do not use it for everything. They know when a mistake could cause harm, when private data is involved and when a qualified human should take the lead.
When accuracy is non-negotiable
Do not rely on AI alone when the cost of being wrong is too high. It may help you prepare questions or understand general information, but a qualified professional should confirm the final answer.
- Medical advice. AI may help someone prepare questions or understand general information, but diagnosis, treatment and medication decisions should be discussed with a qualified healthcare professional.
- Legal advice. In 2025, legal representatives submitted fabricated case citations to the English High Court. Use a qualified legal professional for anything binding.
- Financial decisions. FCA research shows that some consumers use AI for financial guidance while remaining concerned about accuracy and their protection if something goes wrong. Do not rely on generated information as personal investment, financial or tax advice.
- Safety-critical decisions. Follow expert guidance where a mistake could cause physical harm.
- Current specialist facts. Use an up-to-date and authoritative source when the information changes quickly.
When private data is involved
Be careful about what you paste into an AI tool. Public tools may store or process information in ways that are not suitable for sensitive personal or business data.
- Personal information. Remove names, addresses, medical histories and identifying details.
- Financial records. Do not paste bank details, account numbers, salaries or tax records.
- Customer data. Do not use public AI tools with client information without a lawful and approved process.
- Business secrets. Keep confidential strategies, research and internal documents out of public tools.
- Passwords and codes. Never share login details, recovery codes or authentication information.
When human judgement is irreplaceable
- Hiring and people decisions. AI may support administration, but people should make accountable decisions about people.
- Personal advice. Career, relationship and life decisions benefit from someone who knows your real situation.
- Authentic creative choices. Your voice, values and point of view should remain yours.
- Vulnerable people. Children, older people and people in crisis need human care and responsibility.
- Conflict and negotiation. AI can help draft words, but it cannot carry the relationship or consequences.
When a hallucination would be too risky
- Contracts and agreements. A lawyer should review any wording that may become legally binding.
- Production code. Security-critical and high-reliability code needs expert testing and review.
- Academic work. Verify every claim and citation, and follow the rules of your institution.
- Public communications. A person should review anything published in the name of a business or organisation.
- Accountable decisions. Do not use AI to distance yourself from a result you are responsible for.
A quick decision checklist
- Will I fact-check it? If not, do not use AI for anything important.
- Am I sharing sensitive data? If yes, stop and use an approved process.
- Are the stakes high? Use AI only to support qualified human judgement.
- Does this need a human touch? Empathy, authenticity and responsibility cannot be automated away.
- Can I take responsibility? Only use the final output if you are willing to stand behind it.
Prepared by AI Did This
Last reviewed: 28 July 2026
Sources
- Courts and Tribunals Judiciary: Ayinde v London Borough of Haringey and Al-Haroun v Qatar National Bank (opens in a new tab) (6 June 2025)
The official judgment supporting the legal-advice caution.
- Financial Conduct Authority: The Mills Review: AI and the future of retail financial services (opens in a new tab) (6 July 2026)
Research on consumer use, trust, accountability and protection in AI-assisted financial decisions.
- UK National Cyber Security Centre: ChatGPT and large language models: what's the risk? (opens in a new tab) (14 March 2023)
Supports the caution against entering sensitive information into public AI tools.
- Information Commissioner's Office: Guidance on AI and data protection (opens in a new tab)
Guidance relevant to handling personal and customer data.