#40: OpenAI just made the case for FIDA. Is the EU listening?
OpenAI just made the case for FIDA. Is the EU listening?
Cambridge CCAF Global AI Report · AI Act 2027 deadline · OpenAI x Plaid · FIDA off the EU roadmap · UK Open Banking · BIS on financial health · Agentic AI everywhere
A quick personal note before we dive in. The Cambridge Centre for Alternative Finance has published the 2026 Global AI in Financial Services Report. I had the privilege of being on the research team behind it. The report captures how AI is actually being adopted across the sector and how it is landing on providers, vendors, regulators and the consumers on the receiving end. Many thanks to the whole team and to everyone in my insurance and regulatory network who took time to share their perspective with the researchers. You shaped what is in there. This is a starting point for me, not a finish line. The questions that interest me most sit at the supervisory end of the picture. That is where I want to take the writing in the coming months. Read more in Article 01.
Two other updates worth sharing. I have been actively building a network of consultants across Estonia and abroad so we can take on larger development projects for governments and supervisors. The themes are familiar to readers of this newsletter: open finance and insurance, SupTech, fintech strategy and market mapping, digital consumer protection and alignment with EU digital finance and insurance regulations. If your institution is scoping work in any of these areas, get in touch.
I have also been speaking about LinkedIn as a tool for B2B sales, building an international consulting business and supporting professional growth. Two universities, two conferences, one private workshop for a consulting firm. The response has been better than I expected. I am now considering teaming up with one of the best LinkedIn trainers around to build a no-fluff, practical online course for financial and insurance professionals. Would something like that be useful to you? Hit reply and tell me.
The 2026 Global AI in Financial Services Report is out
The Cambridge Centre for Alternative Finance (CCAF) has published the 2026 Global AI in Financial Services Report. The research is based on 628 organisations across 151 jurisdictions: 203 fintechs, 149 financial incumbents, 146 AI vendors and 130 central banks and other financial regulators. It is the first global cross-segment view of how AI is actually being adopted, what it is doing to providers and to supervisors and where the risks are landing.
A few findings worth sitting with:
- 81 per cent of surveyed financial services firms are adopting AI at some level. 40 per cent of industry respondents report advanced adoption (scaling or transforming), more than double the share among regulators (20 per cent).
- Agentic AI is already in active adoption among 52 per cent of industry respondents. 23 per cent are at more mature stages, 29 per cent are piloting.
- Four of the top five industry AI use cases are back-office functions: process automation (79 per cent), data visualisation (75 per cent), software engineering (75 per cent) and data and knowledge management (69 per cent).
- AI is mainly being used to improve execution, not to reconfigure business models. Although 51 per cent of more mature adopters are piloting or deploying new AI-powered products, against 28 per cent among less mature institutions.
- Only 40 per cent of respondents report increased profitability from AI. 43 per cent report no change. Higher spend is strongly associated with greater impact: 62 per cent of organisations spending over USD 100,000 annually on AI have reached advanced maturity.
- The top risks identified are model hallucinations, unreliable outputs, model opacity, weak explainability and market abuse.
Read the full report at Cambridge Judge Business School.
The AI Act now has a binding 2027 deadline. Use it.
On 7 May 2026 the Council presidency and the European Parliament reached a provisional agreement to streamline parts of the EU AI Act as part of the Omnibus VII simplification package. The headline most relevant for the financial sector: the application date for high-risk rules is now fixed at 2 December 2027 for stand-alone high-risk AI systems and 2 August 2028 for high-risk AI systems embedded in products. The deadline for national regulatory sandboxes is also pushed to 2 August 2027.
Two consequences for our sector are concrete. For life and health insurance, AI systems intended for risk assessment and pricing in relation to natural persons are classified as high-risk under the AI Act. For banking, AI systems intended to evaluate the creditworthiness of natural persons or to establish their credit score, with the exception of AI systems used for detecting financial fraud, are equally classified as high-risk.
In other words, the most consequential AI deployments at the heart of the financial sector business model now have a binding deadline. The deadline moved because the technical standards and supporting tools are not ready, not because the substance of the rules became less important. Anyone reading this as permission to slow down is misreading the moment.
Two years is a long time to do nothing and a short time to do this right.
Council press release: Council and Parliament agree to simplify AI rules.
OpenAI just made the case for FIDA
On 15 May OpenAI launched a personal finance experience inside ChatGPT, in preview for US Pro subscribers. Users can connect bank accounts through Plaid, with support for more than 12,000 institutions. Spending, subscriptions, investments and liabilities now flow into a dashboard inside the chatbot. OpenAI has flagged Intuit integration next, which would extend the picture to tax. The company has previously said roughly 200 million people ask ChatGPT financial questions every month. Those questions are about to be answered against real balances and real transactions.
Now consider what this means for Europe.
Plaid works in the United States because it has built private commercial agreements with thousands of banks. That model does not replicate here. Twenty-seven member states, dozens of incumbents per market, no single aggregator with that reach. Without the EU’s Financial Data Access Regulation (FIDA) mandating standardised APIs and common data schemas across banking, insurance, pensions and investments, an AI assistant operating in the EU would need to negotiate access country by country, institution by institution, product by product. It does not scale.
FIDA is the precondition for AI-mediated financial services in Europe. The OpenAI launch is the clearest argument yet for why standardisation cannot be left to the market. If I were a fintech or a financial services incumbent in Europe right now, I would be knocking on the door of the Council and the European Parliament asking for FIDA to be finalised on a fast track.
OpenAI: A new personal finance experience in ChatGPT.
One Europe, one market, no FIDA
On 24 April 2026 the Presidents of the Council, the European Parliament and the European Commission signed the “One Europe, One Market” Roadmap, a priority list of legislative files the institutions commit to agree by end 2027. Neither FIDA nor the Retail Investment Strategy is on it. Three pieces of news this month should be read together.
Calling something “One Europe, One Market” while leaving FIDA off it is a choice. I cannot say I am surprised. I can say I am sad.
Meanwhile in London, the FCA published KPMG’s independent assessment of the two candidates that proposed to lead the standards-setting body capable of becoming the Future Entity for UK Open Banking. An industry panel scored both proposals across six categories covering governance, industry support, technical capability, funding, risk and security as well as operational feasibility. One proposal came out clearly ahead on every category. A second KPMG report on operationalisation is due in the coming weeks. Meaningful progress on establishing the Future Entity is expected by end 2026.
And in Strasbourg, the Lídia Pereira report on financial literacy and finfluencers was adopted. It calls for AI-powered personalised financial education, an EU comparison mechanism for Savings and Investment Accounts, low-cost cross-border SIA portability and a clear-eyed treatment of finfluencers, finchat and the deepfake economy. Insurance and pensions, often forgotten in the literacy debate, are explicitly mentioned. None of these ambitions work without FIDA. Personalised tools, real comparison and real portability all need data flowing across providers in a structured way. That is exactly what open finance and FIDA are meant to enable.
Open finance, AI and a strong consumer protection framework are not three separate policy debates. They are the same debate. Get the combination right and we address structural problems in retail finance and insurance we have been describing for a decade. Recent AI developments make FIDA more important than ever, not less. Especially for those who fight against it the hardest.
What exactly are we afraid of?
Sources: One Europe, One Market Roadmap (Council) · FCA independent assessment of Open Banking Future Entity proposals · Pereira report A10-0082/2026 (European Parliament)
Innovation is necessary, but not sufficient
Financial Stability Institute (FSI) Brief No. 31 from the Bank for International Settlements puts the consumer-outcome question in plain language. Digital innovation is enhancing access to payments, credit, savings and insurance. It can help people manage their financial obligations and have greater confidence in their financial future. But these benefits are emerging alongside new vulnerabilities: a global surge in scams and fraud, greater overindebtedness among some digital borrowers and the use of ill-suited investment products. Aggregate trends in financial health are mixed. In some countries the available indices are deteriorating despite greater uptake of digital technologies.
The insurance angle is what caught my eye. The brief notes that gradual extension of open finance frameworks to insurance may further support innovation, including through consent-based data sharing to improve underwriting and product tailoring. Early developments in jurisdictions such as India and Brazil suggest that open finance ecosystems may increasingly enable more integrated insurance offerings and help expand coverage to individuals and small firms with limited traditional financial histories. Digitalisation could support more inclusive and scalable insurance markets, provided appropriate consumer protection safeguards are in place to ensure transparency, fair pricing and suitable product design for underserved groups.
Innovation is necessary but not sufficient. The supervisory framework around it will decide whether it delivers. Curious how this is landing in your jurisdiction. BIS FSI Brief 31: Digitalisation and innovation, opportunities and risks for financial health.
Industry call to redesign insurers as “expert-centric” for the agentic AI era. Four building blocks: leadership, human experts, synthetic execution, orchestration managers. Only 10 per cent of P&C insurers are successfully scaling AI; 42 per cent track no AI metrics. Capgemini’s top performers, which it calls “intelligence trailblazers,” have outperformed their peers by up to 21 per cent in revenue growth over a three-year period. Capgemini.
Templates for KYC screening, month-end close, valuation review, statement audit and pitchbook building. Verisk is now a connector for property, casualty and specialty data flowing into agentic workflows for underwriting, claims and risk. Travelers is among the named customers. The shift is not the AI itself, it is the packaging: security, audit trails and controls already built in. The gap between firms experimenting in pilots and firms running this in production is about to widen fast. Anthropic.
Joint paper from the CMA, FCA, ICO and Ofcom. The framing matters: regulation as an enabler of innovation, not a brake on it. If an AI assistant recommends a financial product, the FCA’s remit applies. Consumer Duty applies too. Agentic AI does not respect borders. The questions the UK is asking today are the same ones the EU needs to answer. DRCF.
One concrete insurer’s perspective on managing AI risk. The dual role of insurers, deploying AI internally while underwriting AI risk for customers. The AI risk spectrum, from bias and unsafe outputs to opacity and accountability gaps to systemic risk from concentration in foundation models. The insurability question: you cannot insure what you cannot understand, quantify or attribute. Policy asks: clearer liability, AI incident reporting modelled on cyber, concentration risk, workforce transition. As the paper puts it, the goal is not zero risk, it is risk that is understood, governed and proportionate to impact. Zurich.
What if AI models are starting to operate so quickly and autonomously that they shape financial market behaviour in ways investors do not always realise? The Dutch supervisor is opening the conversation with the market and with national and international supervisors. The right order: dialogue first, deployment second. AFM.
The FCA has named general insurance an explicit innovation priority for 2026. Applications to its Innovation Services grew 49 per cent in 2025, but only 1 per cent of 2025 applications came from general insurance and protection, against 35 per cent from payments and digital assets. After years of innovation policy skewing toward payments, banking and wealthtech, insurance is being brought back into the centre of the sandbox conversation. Why do we still see fewer insurers in sandboxes? FCA.
Anyone still talking about the metaverse? Didn’t think so. The hype died, the headlines moved to AI. Easy to assume the whole thing is over. It isn’t. The European Commission’s Joint Research Centre mapped 88,000 virtual world activities across 68,000 players globally. The EU shows rising virtual world activity in manufacturing and in financial and insurance activities. European Commission.
Total fintech revenues in 2025, representing only 4 per cent penetration of the wider financial services market, according to the new McKinsey and QED Investors report “The Next Age of Fintech.” The report projects fintech revenues to triple to USD 2 trillion by 2030 if recent growth holds. McKinsey & QED Investors.
Weekly briefing on financial innovation and regulation. Join 4,700+ fintech, insurance and regulatory professionals.
SubscribeAdvisory for regulators, boards and fintech leaders navigating digital finance policy and regulation. See how I can help.
Selectively considering sponsorship for this newsletter. Reach 4,700+ decision-makers in financial innovation and regulation. Enquire.
