FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: Actuaries
UK actuaries using AI must meet TAS 100 and the Actuaries' Code, while the FRC Models guidance and the IFoA data science and AI guide are persuasive rather than mandatory. Knowing which is which shapes how a regulator, a court or an insurer is likely to judge an AI-related error.
Part of: AI in Actuaries
In short
There is no AI-specific rulebook for UK actuaries. The binding requirements are the FRC's TAS 100, which the FRC says applies to AI and machine learning models used in actuarial practice, and the IFoA's Actuaries' Code and Actuarial Profession Standards. On top of that sit non-mandatory documents: the FRC Technical Actuarial Guidance: Models (October 2024), the IFoA AI risk alert (September 2023) and the IFoA ethical and professional guide on data science and AI (V2.1, July 2025). None of them says anything about PI insurance, but they set the standard of care an actuary's work is likely to be measured against.
Two bodies matter. The Financial Reporting Council sets the Technical Actuarial Standards (TASs), oversees the IFoA's regulation of its members and runs the FRC Actuarial Scheme for enforcement (FRC actuarial standards). The Institute and Faculty of Actuaries sets the Actuaries' Code and the Actuarial Profession Standards (APSs) and runs an independent disciplinary process (IFoA standards).
Work-specific standards also apply: TAS 200 for insurance work, TAS 300 for pensions, and IFoA standards such as APS P1 (pension schemes), APS L1 (life assurance), APS G1 (Chief Actuary, non-life) and APS X3 (expert witness work). None of these was written for AI, but each still governs work that uses it.
TAS 100 v2.0, "General Actuarial Standards", was published on 3 March 2023 and took effect on 1 July 2023 (TAS 100). The FRC states that TAS 100 is applicable to AI and machine learning models used in actuarial practice. The principles AI touches most are:
The Actuaries' Code (v3.2, effective 25 September 2025) has six principles: integrity, competence and care, impartiality, compliance, speaking up and communication (the Actuaries' Code). The 2025 revision concerned respect and speaking up against unfair treatment and does not mention AI. APS X2 v1.1, in force from 30 January 2026, requires members to consider whether Work Review and Independent Peer Review are appropriate; it also does not mention AI.
Published on 7 October 2024, the Models guidance supports TAS 100 Principle 5. The FRC describes it as "persuasive not prescriptive" (1.1). Key points:
Generative AI is mentioned only in passing, with a note that different steps may be required.
The IFoA issued an AI risk alert in September 2023. As recorded in its later guide (2.45), the alert warns that "the use of AI in actuarial work may amplify existing risks and introduce new risks."
The main IFoA document is the Ethical and professional guidance on Data Science and Artificial Intelligence (IFoA Regulatory Board, V2.1, July 2025). It began in 2021 as data science guidance and was extended to wider AI in 2024. It states that "it is not compulsory for Members to comply". It maps AI issues to all six Code principles and covers clear governance and agreed responsibility in multidisciplinary teams (2.11, 2.29), confidentiality and consent (2.10 to 2.11, 2.20), bias (2.17), readily interpretable outputs (2.15), care with third-party generative AI tools (2.14 to 2.15), and peer review under APS X2 with caution about using AI within reviews (2.35 to 2.38).
Supporting material includes the 2019 RSS and IFoA ethical data science guide, a 2025 CPD reflective toolkit on data science and AI, and a partnership with CISI (22 July 2024) giving members access to an ethical AI certificate.
The FRC's October 2023 research, carried out by the Government Actuary's Department through 37 interviews and more than 100 survey responses, found many actuaries planned to increase AI use, especially large language models, with general insurance pricing the heaviest current user. The Lloyd's Market Association's 2025 discussion paper on AI and the Lloyd's actuary is not a standard, but recommends model ID cards and version control.
| Document | Body | Date | Status |
|---|---|---|---|
| TAS 100 v2.0 | FRC | Effective 1 July 2023 | Mandatory |
| Actuaries' Code v3.2 | IFoA | Effective 25 September 2025 | Mandatory |
| APS X2 v1.1 | IFoA | In force 30 January 2026 | Mandatory |
| Technical Actuarial Guidance: Models | FRC | 7 October 2024 | Guidance, persuasive |
| AI risk alert | IFoA | September 2023 | Risk alert |
| Data science and AI guide V2.1 | IFoA | July 2025 | Guidance, not compulsory |
| AI and the Lloyd's actuary | LMA | 2025 | Discussion paper |
Non-mandatory does not mean irrelevant. If an AI-assisted piece of work goes wrong, a disciplinary panel, an expert or a court may look at the FRC and IFoA guidance as evidence of what a competent actuary would have done. Following it, and recording why, is a practical defence; departing from it without a reason is hard to explain.
No regulator or IFoA statement on AI and PI insurance has been found. We have not found an IFoA rule requiring members to hold PI insurance; APS X5 is titled "Compensation for Professional Shortcomings", so check its current text. FCA-authorised actuarial firms follow the FCA's PI rules for their permission type. See actuaries and PI: IFoA requirements.
How the guidance plays out by task: machine learning pricing models, relying on third-party models and peer review under APS X2. For cover, see whether actuarial PI covers AI mistakes, and the AI in actuarial work overview.
Open questions include whether the FRC will update its guidance specifically for generative AI, how its October 2025 guidance on proportionality is applied to AI model governance, and any update to the IFoA's 2023 risk alert. The FRC's Actuarial Monitoring Programme has not yet published AI-specific findings.
Apex arranges PI for actuaries and actuarial consultancies; a broker can review how your policy treats AI-assisted work.
No. The FRC describes it as persuasive not prescriptive. It explains how to meet TAS 100 Principle 5, which is mandatory, so departing from the guidance without good reason makes it harder to show that TAS 100 was met for an AI or machine learning model.
No. The Code (v3.2, effective 25 September 2025) does not mention AI, but its principles of competence and care and of communication apply to any work that uses AI. The IFoA data science and AI guide maps AI issues to each of the six principles.
Only lightly. The FRC Models guidance mentions generative AI in passing. The IFoA guide says care should be taken when using third-party generative AI tools, citing veracity, privacy and copyright. Whether the FRC will issue specific generative AI guidance is an open question.
Not directly; none of it addresses insurance. In practice, insurers may ask how you govern AI models, and the guidance gives a ready framework for that answer. Following it also helps defend a negligence claim, which matters to you and your insurer.
Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.
Start a PI quote Request a callbackApex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.
