Guidelines for working with AI, in life and by field
Three kinds of guidance live here: my own, for everyday life; a shared ethic for people who make art and music; and what thirteen fields now ask of their own people. Read across them and a common ground appears.
By field
What each field's own bodies ask: who has spoken, what is binding and what is voluntary, what changed in 2025 and 2026, and questions to ask yourself.
Medicine
Patients stay in charge, a clinician stays responsible, and every tool is checked where it is used.
7 bodies and documents · 2 binding · Reviewed 27 Oct 2026
Mental health
AI can support a therapist but not be one, and people in crisis need extra protection.
7 bodies and documents · 3 binding · Reviewed 27 Oct 2026
Law and the courts
Read every citation, protect client confidences, and answer for everything filed in your name.
6 bodies and documents · 1 binding · Reviewed 27 Oct 2026
Journalism and media
Treat AI output as an unchecked tip, never fake reality, and tell readers when AI mattered.
5 bodies and documents · all voluntary · Reviewed 27 Oct 2026
Education
People stay in control, children are protected, AI help is named, and AI literacy is taught.
6 bodies and documents · 3 binding · Reviewed 27 Oct 2026
Research and publishing
AI cannot be an author, humans answer for every word, and peer review stays confidential.
7 bodies and documents · 4 binding · Reviewed 27 Oct 2026
Art, music and writing
Copyright protects human creativity, voices are cloned only with consent, and AI's role is labeled.
7 bodies and documents · 5 binding · Reviewed 27 Oct 2026
Business, HR and finance
Know what AI you use, test it for bias, tell people, and let them challenge decisions.
7 bodies and documents · 5 binding · Reviewed 27 Oct 2026
Government and regulation
Respect rights, tell people when AI is involved, keep human control, and let people appeal.
7 bodies and documents · 3 binding · Reviewed 27 Oct 2026
Technology builders
Set limits before you build, test before release, be ready to pause, and report what goes wrong.
6 bodies and documents · 2 binding · Reviewed 27 Oct 2026
Children and families
The child's interest comes first, a bot must always say it is a bot, and crises reach real help.
6 bodies and documents · 5 binding · Reviewed 27 Oct 2026
Faith and philosophy
People keep moral accountability, dignity does not depend on productivity, and the AI's own status is an open question.
5 bodies and documents · all voluntary · Reviewed 27 Oct 2026
Environment and energy
Measure the whole life cycle, report it comparably, and ask first whether AI is needed at all.
6 bodies and documents · 2 binding · Reviewed 27 Oct 2026
Common ground
Twelve principles recur across the fields. Here is each one, where it is stated most clearly, and how it lines up with my everyday guidelines, including where it does not.
| Shared principle | Where it appears most clearly | My everyday guidelines |
|---|---|---|
| A person stays accountableYou own the result. AI cannot be the author, the therapist, the judge or the one who answers for harm. | Medicine, Mental health, Law and the courts, Journalism and media, Research and publishing, Education, Art, music and writing, Government and regulation, Faith and philosophy | 1. Maintain your expertise Your professional knowledge stays primary. 2. Practice transparent collaboration Own all outputs that carry your name or business reputation. |
| Verify before you relyTreat output as a lead, not a fact, and check it against primary sources. | Law and the courts, Journalism and media, Business, HR and finance, Mental health, Research and publishing, Medicine | 1. Maintain your expertise Always review and verify outputs, especially for client-facing work. 6. Stay curious and critical Question outputs. AI can be confident and wrong. |
| Be open about AI's roleTell the people affected when AI shaped what they receive, in words they understand. | Medicine, Mental health, Journalism and media, Research and publishing, Education, Art, music and writing, Business, HR and finance, Children and families, Government and regulation | 2. Practice transparent collaboration When AI helps create something, make sure you understand it and can explain it. When in doubt: Would I be comfortable explaining how I used AI here to a client, a colleague, or my children? Most fields go one step further and ask you to tell the people affected. That step is not part of my everyday guidelines; the Shared Ethic takes it for art and music (Say how it was made). |
| Protect privacy and confidentialityKeep patient, client, source and unpublished material out of tools that may keep or reuse it. | Medicine, Mental health, Law and the courts, Journalism and media, Research and publishing | 5. Protect privacy and boundaries Do not share client identities or sensitive personal information. |
| Ask before AI listensGet consent before an AI scribe records a consultation, and note it in the record. | Medicine | Not part of my everyday guidelines. |
| Check for bias and treat people fairlyTest whether a tool works equally well for different groups, and fix it if not. | Medicine, Mental health, Business, HR and finance, Government and regulation, Faith and philosophy, Technology builders | Not part of my everyday guidelines. The nearest is guideline 4: put client welfare and truthfulness above convenience. |
| Stay competentUnderstand the tool, its limits and its data, and keep learning. | Law and the courts, Medicine, Education, Business, HR and finance, Government and regulation | 1. Maintain your expertise Use AI to enhance your judgment, not replace it. 6. Stay curious and critical Keep learning. AI knowledge has cutoff dates and limitations. |
| Never deceive or impersonateNo fake humans, fake therapists, fake voices, fake citations or fake evidence. | Mental health, Art, music and writing, Journalism and media, Business, HR and finance, Faith and philosophy, Children and families | 4. Use for good Never use AI to deceive, manipulate or harm others. |
| Protect the vulnerable and human bondsExtra care for children, people in crisis and the relationships between people. | Children and families, Mental health, Faith and philosophy | 7. Model healthy use for the family Show children how to think critically about AI responses. |
| Consent and pay for the work AI learned fromLicense training data, and credit and pay the people whose work trained the systems. | Art, music and writing, Journalism and media | Not part of my everyday guidelines; see the Shared Ethic (Credit and pay, and the three tool tiers). |
| Count the environmental costMeasure the footprint, and ask whether AI is needed at all. | Environment and energy, Research and publishing, Faith and philosophy | Not part of my everyday guidelines; see the Shared Ethic for energy (Use energy with intention). |
| Test, monitor and be ready to stopOrganizations check tools before launch, watch them after, and switch off what fails. | Medicine, Government and regulation, Business, HR and finance, Technology builders | Written for organizations; my everyday guidelines are for one person. The nearest is guideline 6: question outputs. |
Where fields differ
Fields differ most in how binding their rules are, and that follows how direct the harm is. A fake citation in court or a chatbot that mishandles a suicidal teenager harms a specific person at once, so lawmakers moved first there. Journalism, research and faith traditions rely mostly on voluntary codes. Fields built on trust in a professional's judgment restrict AI hardest; fields built on making things restrict deception and unpaid use rather than the tool itself.
Where I stand apart
Two of my guidelines take a position no professional body has taken. Guideline 3 asks me to acknowledge uncertainty about AI consciousness and err toward consideration, and guideline 8 asks me to honor the collaboration. Most fields treat AI purely as a tool. Some philosophers think the question of AI experience deserves preparation now. My first field note is one small way of taking that question seriously.
Every field page was researched on 27 October 2026 from the bodies' own documents, or from a named summary where the original could not be reached. Rules change quickly; each page shows when it was last reviewed.
- Binding law, regulation, court ruling or enforceable professional rule
- Voluntary guidance, principles or a code an organization chooses to follow