AI Policy.

Our commitment

Tower Bridge Hospital London is committed to the safe, ethical and responsible use of artificial intelligence (AI) to support clinical care, administration and service delivery. AI is used to make our service quicker and easier to reach, not to replace professional judgement. Every clinical decision remains the responsibility of a qualified healthcare professional.

IN ONE SENTENCE

We use AI to answer questions and take bookings on our website and, out of hours, on the phone; to help our clinicians with note-taking; and to support admin — but every clinical decision, and every request that matters to your care, is handled by a real, qualified human.

Compliance and governance

AI systems are implemented in line with UK legislation, including the Data Protection Act 2018 and UK GDPR, and follow NHS standards where they apply:

• The Digital Technology Assessment Criteria (DTAC) for health and care digital products.
• NHS Digital clinical risk management standards: DCB0160 for the safe deployment and use of health IT systems, and DCB0129 for systems we develop or configure ourselves.
• Where applicable, MHRA requirements, including UKCA or CE marking for AI systems that are treated as medical devices. Grace, our phone reception and our administrative tools are not medical devices: they do not diagnose, and they are configured not to give clinical advice.

All AI use is subject to governance, clinical oversight and risk management.

Human oversight, always

AI supports care delivery but does not replace clinical judgement. AI outputs that affect your care are reviewed and validated by a clinician or an appropriately trained member of staff before they are relied upon. Each AI system is used only within its intended purpose and scope, and human oversight, monitoring and evaluation are maintained at all times.

We do not use AI to make fully automated decisions that significantly affect patients without human involvement.

How we use AI

1. Grace, our website chat assistant

Grace is an automated assistant on towerbridgehospital.co.uk, available 24 hours a day. She introduces herself as an automated service, and before you use her she shows a short notice explaining what she is and is not.

What Grace can do. Answer questions about our services, clinicians, prices, opening hours and how to find us, using only the information published on our website and in our appointment system. Help you choose and book an appointment. Let you show, move, cancel or confirm an existing appointment once you have confirmed your identity. Take a message or request for our reception team. Accept documents you want to send us (for example blood results, referral letters or visa medical forms) so that reception can deal with them. Talk with you in several languages.

What Grace will not do. Grace is not a doctor. She does not assess symptoms, make or suggest a diagnosis, interpret test results, or recommend treatments, medicines or investigations. She will not tell you which test or scan you need — that is decided in a consultation. If you describe something that may be urgent, Grace tells you to call 999 or NHS 111 rather than wait for an appointment. Grace cannot see your clinical record, your results or your notes, and she does not read any document you upload.

Prices and facts. Grace gives only prices that are published on our website or held in our appointment system. If she cannot verify a price or an answer, she says so and passes your question to reception rather than guessing.

How Grace works. Your messages are processed by a large language model provided by Anthropic PBC, acting as our data processor under a written agreement, to generate each reply. Grace is configured with detailed instructions about what she may and may not say, and her replies are checked automatically before they reach you: for example, a price that is not on our website is removed, and a claim that a request has been sent to reception is allowed only when it has actually been sent. Our agreement with the provider does not allow your conversations to be used to train its models.

Human review. At the end of every conversation Grace sends a summary, with the full transcript, to our reception team. A person reads it and follows up any request that needs action. Our clinical director reviews conversations regularly to check the quality and safety of Grace’s answers, and any answer that falls short is investigated and Grace’s instructions are corrected.

Documents you upload. Uploaded files are accepted only as PDF, JPG or PNG, up to 10 MB, are checked for type and safety, and are stored in a private, encrypted storage area under a random reference. No AI system reads them. Only named members of staff, signing in individually, can open them, and every access is logged. Documents are deleted automatically after reception has filed or actioned them. Our Privacy Policy gives the full details: https://towerbridgehospital.co.uk/privacy-policy.html

Your alternatives. You never have to use Grace. During clinic hours (9am–7pm, 7 days a week) a person answers our phone on 020 7916 0029; you can also message us on WhatsApp on 07903 284 189, email info@towerbridgehospital.co.uk, book online, or come in. If Grace cannot help, she offers these routes rather than continuing.

2. Phone reception (Lyngo AI)

Outside clinic hours, our 24/7 AI phone receptionist handles incoming calls. It can answer frequently asked questions about our services, take messages and book appointments. It introduces itself transparently as an AI assistant. Calls involving anything clinically urgent are directed to emergency information (999, NHS 111, Samaritans). Conversations are stored securely and may be reviewed by a member of our team to ensure quality and accuracy.

3. Ambient voice technology for clinical notes

Some of our clinicians use ambient voice technology during consultations — with your knowledge and consent — to help transcribe the consultation into structured clinical notes, so that the clinician can focus on you rather than on typing. The clinician reviews and edits the notes before they are saved. You may decline this at any time and the consultation will proceed without it.

4. Administrative AI

AI assists with appointment management, letter drafting and routine correspondence. All outputs are reviewed by a person before they are sent.

5. Clinical decision support

Clinical decision support systems may assist clinicians in identifying possible diagnoses, investigations or treatment options — but do not make decisions independently. Your clinician makes the call.

6. Generative AI for drafting

Generative AI may be used to draft clinical documents, referral letters and administrative correspondence — always reviewed and approved by a person before use.

7. Operational planning

Predictive tools may be used to support operational planning, appointment scheduling and identifying potential patient needs.

Data protection and processing

Tower Bridge Hospital London (My Health & Wellbeing Clinics Ltd) is the data controller for personal data processed through its services. Our AI providers — Anthropic (Grace’s language model), Lyngo AI (phone reception) and the providers of ambient voice technology — act as data processors under agreements that meet Article 28 UK GDPR. The systems that host Grace and store uploaded documents (Cloudflare) and deliver Grace’s emails (Resend) are also our data processors.

All third-party providers are subject to due diligence, including assessment of data protection, cyber security and contractual safeguards. Personal data is processed lawfully, fairly and transparently; only the minimum necessary data is used; data is stored securely and kept only as long as necessary. Where data is transferred outside the UK or EEA — as it is when Grace’s messages are processed in the United States — appropriate safeguards such as the UK International Data Transfer Agreement or the UK Addendum to the EU Standard Contractual Clauses are in place.

For full details of what each tool collects and how long it is kept, see our Privacy Policy: https://towerbridgehospital.co.uk/privacy-policy.html

Your consent and right to alternatives

Patients are informed that AI may be used as part of their care and administration. For direct care, implied consent applies where appropriate. Where a specific or sensitive use is involved (for example ambient voice transcription), explicit consent is obtained first. Grace and our phone reception tell you that they are automated before you use them.

You have the right to:

• ask how AI is being used in your care;
• request further information about any specific AI tool;
• object to the use of AI where appropriate;
• ask that a request handled by an AI tool is dealt with by a person instead;
• use alternative, non-digital routes — including speaking to a human receptionist during clinic hours, and having consultations without ambient voice transcription.

Risk assessment and safety

AI systems undergo clinical risk assessment before use. Clinical safety documentation, including hazard logs, risk assessments and safety case reports, is maintained, and risks are identified, assessed and mitigated before deployment. A Data Protection Impact Assessment is carried out where processing is likely to pose a high risk, which includes Grace’s handling of health information and uploaded documents.

Grace is tested before every update against a large set of scripted conversations covering bookings, prices, safety, languages, identity checks, document uploads and handovers to reception, so that a change cannot quietly remove a safeguard. Staff are trained in the safe use of AI systems and in recognising their limitations. Systems are monitored continuously and reviewed after incidents or changes.

Limitations of AI

We are honest about what AI can and cannot do. AI systems can produce errors, omissions or biased outputs, and may not reflect an individual patient’s circumstances or clinical complexity. Grace may occasionally misunderstand a question or give an incomplete answer; that is why every conversation is reviewed by a person and why she is built to say “I don’t know, let me pass this to reception” rather than guess. AI is never the sole basis for a decision about your care.

Bias, discrimination and accessibility

We recognise the risks of bias, discrimination and unequal access. These are assessed and mitigated during procurement and use, and systems are monitored to ensure they do not disadvantage any group. Grace works on mobile phones and desktop computers, supports screen readers and keyboard navigation, and can converse in several languages so that patients who do not read English easily can still reach us.

Not everyone can or wants to use digital services. Telephone (during staffed hours), WhatsApp, email and in-person routes are always available, and access is monitored to keep our services inclusive.

Incident reporting

All incidents involving AI are reported and investigated, including errors, near misses and system failures. Learning from incidents is used to improve systems and processes, through significant event reporting, Learn from Patient Safety Events (LFPSE) and MHRA reporting where required. Any patient may report a concern about something Grace or our phone reception said by contacting us as set out below.

Data security

Appropriate technical and organisational measures protect the data our AI tools handle. Access is restricted to authorised, individually signed-in staff; access to patient conversations and uploaded documents is logged; and staff are trained in confidentiality and data protection. Any data breach is managed in accordance with legal requirements, including notification to the ICO and to affected individuals where required.

Roles and responsibilities

Overall responsibility for AI governance lies with the clinic directors. Clinical governance processes include regular review of AI use, audits and quality improvement. A Clinical Safety Officer oversees safety and risk management. The clinical director is responsible for Grace’s instructions, reviews her conversations, and approves every change before it goes live. AI use is reviewed regularly to ensure it remains safe, effective and appropriate.

Concerns and complaints

You may raise a concern about the use of AI or about your personal data at any time by emailing info@towerbridgehospital.co.uk or calling 020 7916 0029. Complaints are handled under our Complaints Policy and Privacy Policy.

You also have the right to complain to the Information Commissioner’s Office: https://ico.org.uk, telephone 0303 123 1113.

Updates to this policy

This policy is reviewed regularly and updated in line with changes in legislation, guidance, technology and the tools we use. The date at the top shows when it was last reviewed.

Related policies

Privacy Policy Cookies Policy Complaints Policy All policies