Theory of Change

The problem

Artificial intelligence is being deployed at scale across healthcare, education, financial services, and the wider public sector, but the United Kingdom has, to date, declined to introduce comprehensive statutory safeguards. There is no domestic legislation imposing binding transparency obligations on AI developers; no independent audit framework; no statutory duty of disclosure as to training data, system instructions, or commercial arrangements influencing model outputs; and no proper redress for individuals subject to opaque automated decisions. Voluntary principles, technical standards, and industry codes of conduct have not closed this gap, and they were never designed to.

Long-term outcome (10-year vision)

The public, the professions, and policymakers in the United Kingdom understand the rights engaged by artificial intelligence and how they can be upheld. The legal and regulatory questions AI raises - transparency, accountability, the protection of children and adults at risk online, the position of human creative work in training data, and meaningful redress where automated decisions go wrong - are well-informed by independent evidence and rigorous analysis. Parliament, the regulators, the courts, and the public are equipped to make better-informed decisions about how AI is developed, deployed, and governed.

Intermediate outcomes (3-5 years)

• Parliamentary inquiries and consultations on AI are routinely informed by FAIR’s evidence, briefings, and analysis.
• The questions raised by AI for copyright and creators’ rights are better understood, and the public conversation is informed by clear, accessible analysis.
• The risks AI presents for online safety, especially for children, are better understood by professionals working in regulated sectors and by the public, and FAIR’s evidence is among the inputs to consultations conducted by Ofcom and government.
• Public AI literacy, supported by FAIR’s plain-language briefings and explainers, is meaningfully improved among Parliamentarians, journalists, and professionals working in regulated sectors.
• FAIR is a recognised, trusted, and independent reference point cited routinely in parliamentary inquiries, regulatory consultations, and serious media coverage.

Activities (what FAIR does)

1. Original research. Evidence-based briefings, working papers, and policy analyses produced to a publishable standard, drawing on peer-reviewed sources, parliamentary material, and primary interviews.
2. Parliamentary engagement. Written and oral evidence to Select Committees, briefings for All-Party Parliamentary Groups, and structured engagement with relevant Parliamentarians.
3. Regulator engagement. Submissions to consultations issued by the ICO, Ofcom, the FCA, the IPO, the EHRC, and other relevant regulators.
4. Public education. Plain-language explainers of policy questions, accessible to professionals across law, healthcare, education, financial services, and the public sector.
5. Convening. Roundtables and working groups bringing together academic, professional, and civil-society voices on specific reform questions.

Inputs (what FAIR needs to do this)

• An expert and independent Board of Trustees.
• Working partnerships with researchers, practitioners, and peer organisations.
• Sustainable, diverse, non-conflicted funding.
• Volunteer and (in time) staff capacity to produce and disseminate research.
• An informed public network of supporters and members.

Assumptions

FAIR’s theory of change rests on three assumptions:

1. That evidence-based, principled engagement with Parliament and the regulators - sustained over years - can shape legislation and regulatory practice. This is the reasonable inference from the role played in earlier reform cycles by, for example, the Bingham Centre for the Rule of Law, JUSTICE, the Howard League, Liberty, and the Ada Lovelace Institute.
2. That artificial intelligence remains, over the period covered by this strategy, a topic on which Parliament is engaged and on which legislative reform is plausible.
3. That the United Kingdom continues to have legitimate, functioning democratic institutions through which the work described above can be conducted.

How we measure progress

• Submissions accepted by Select Committees, with citations or quotation in their reports.
• Substantive references in regulatory consultation responses or final policy statements.
• Citations in peer-reviewed research, government documents, and serious media.
• Membership of FAIR (graded) and engagement with our briefings (downloads, distributions, event attendance).
• Independent annual reporting on progress against this theory of change in our Annual Report.

What we are not

• A think tank for hire. FAIR does not produce work to industry brief.
• A campaigning organisation in the protest sense. We engage Parliament and the regulators by argument, evidence, and convening.
• An accreditation body. FAIR does not certify AI products, models, or vendors.
• A litigation organisation. FAIR may, in time, support strategic litigation conducted by appropriate partners; we do not bring our own cases.