Digital professionalism, messaging, images, cloud systems and AI

Apply confidentiality and data-protection principles to email, messaging, photographs, recordings, remote care, cloud platforms and AI without relying on product-specific rules.

• Assess digital tools by data flow, purpose, approval and safeguards.
• Use secure messaging and email proportionately.
• Manage clinical images and recordings as confidential information.
• Distinguish anonymisation, pseudonymisation and de-identification in digital systems.
• Identify when DPIA, contractual review and human oversight are needed for AI or new technology.

Digital tools do not suspend the ordinary rules of confidentiality. They can increase the speed, scale, persistence and re-identification risk of a disclosure, so purpose, recipient, access control, retention and audit become especially important.

Email and messaging should use organisation-approved systems where required. Before sending, verify recipient, attachment, distribution list and whether the content is necessary. Convenience does not justify using an insecure personal channel for identifiable clinical information.

Clinical photographs, video and audio recordings can be highly identifying and may need explicit consent depending on purpose. Recording for direct care, teaching, publication and personal use are different contexts and should not be collapsed.

Consumer generative-AI services should not be treated as interchangeable with organisation-approved clinical or enterprise AI. A prompt containing patient information may involve disclosure to a provider, international processing, retention, model improvement or other data flows that require governance.

High-risk new technologies may require a DPIA. For significant solely automated decisions, current law requires safeguards including information, representation, human intervention and the ability to contest. Clinical professional accountability also requires meaningful human oversight beyond the minimum legal threshold.

Use CLEAR consistently across information decisions.

1. Clarify the purpose and recipient.
2. Locate the professional and legal authority.
3. Evaluate necessity, proportionality and the minimum information needed.
4. Ask about consent, objections, capacity, statutory duties or powers, and public-interest factors.
5. Record the reasoning, disclosure, advice and safeguards.

CLEAR is an educational scaffold, not a statutory legal test.

Scenario 1. A doctor copies an identifiable discharge summary into a public AI chatbot to improve wording.

Reasoning prompt: clarify purpose and recipient; locate authority; evaluate necessity and minimum information; ask about consent, objections, capacity and risk; record and escalate proportionately.

Scenario 2. A clinician sends wound photographs through a personal messaging app because the approved system is slow.

Reasoning prompt: clarify purpose and recipient; locate authority; evaluate necessity and minimum information; ask about consent, objections, capacity and risk; record and escalate proportionately.

Scenario 3. An organisation deploys an AI triage tool that produces significant decisions with minimal human review.

Reasoning prompt: clarify purpose and recipient; locate authority; evaluate necessity and minimum information; ask about consent, objections, capacity and risk; record and escalate proportionately.

• Removing the name makes an AI prompt anonymous.
• All AI is banned in healthcare.
• An approved AI tool is automatically safe for every purpose.
• If messaging is encrypted, any clinical content is acceptable.

• What exactly is the purpose of using or disclosing the information?
• Who is asking, and have I verified identity and role?
• What professional, common-law, statutory and data-protection authority applies?
• Can the purpose be achieved with less or anonymised information?
• Does the patient expect this use, have they consented or objected, and does that matter in this context?
• Is there a nation-specific rule, local policy or specialist route I need to check?
• What should I document, and who should I consult if uncertain?

Optional reflection: Do not enter identifiable patient, colleague or employer information. Identify one situation in your practice where confidentiality, data protection and local governance could easily be conflated. How would CLEAR improve the reasoning?

• Digital confidentiality uses the same core principles as offline practice.
• Approved systems and local governance matter.
• Verify recipients before sending sensitive information.
• Images and recordings can identify patients even without names.
• Pseudonymisation does not remove data-protection obligations.
• Consumer AI may create external disclosures and retention risks.
• AI deployment should be assessed by data flows and purpose.
• A DPIA may be required for high-risk new technology.
• Significant automated decisions require safeguards.
• Meaningful human oversight is a professional as well as legal issue.

Check current versions for time-sensitive decisions.

• R18 - ICO: Artificial intelligence and data protectionOpen official source
• R16 - ICO: PseudonymisationOpen official source
• R17 - ICO: Anonymisation guidanceOpen official source
• R31 - ICO: Data protection impact assessmentsOpen official source
• R4 - UK Government: Data (Use and Access) Act 2025: plans for commencementOpen official source