Rusu's AI speech: 86% adoption rate reframes the FCA's supervisory posture

Jessica Rusu used a Bloomberg platform on 19 October to put a number on something boards have been sensing for months: 86% of financial services firms now use AI, according to the forthcoming joint FCA and Bank of England survey (FCA). That figure, paired with her confirmation that the FCA's response to the Mills Review will be "testing, assurance, monitoring and surveillance" rather than further consultation, reframes the supervisory contract between the regulator and the regulated.

From rulemaking to field testing

The significance of Rusu's speech lies in what the FCA is choosing not to do. There is no new rulebook. Instead, the regulator is embedding six of the seven Mills Review recommendations into live work immediately, using its AI Lab to see directly what firms are building, from agent-led payments to fraud tools (FCA). This is a deliberate shift. Rather than codifying expectations in advance, the FCA is reserving the right to assess practice in situ and judge firms against principles already on the statute book: consumer duty, senior manager accountability, operational resilience.

For general counsel and chief risk officers, that posture is harder to manage than a prescriptive rulebook. There is no safe harbour of compliance. The regulator's good and poor practice examples, published alongside the speech, will become the reference points against which supervisory conversations are conducted. Firms that cannot articulate how they put consumers first, maintain clear accountability and plan for when things go wrong, in Rusu's own framing (FCA), should expect pointed questions.

The agentic problem is now, not later

Rusu explicitly cites the launch of Meta's Muse as evidence that agentic finance, anticipated by the Mills Review for 2030, is already emerging (FCA). This compresses the planning horizon materially. The governance assumption that firms could watch the retail agent market mature before committing is no longer tenable. The FCA is observing deployments in real time through its innovation services and expects firms to be doing the same internally.

The wholesale read-across matters too. Rusu was clear that the AI shift does not stop at the retail boundary, and that across wholesale markets firms are using AI to analyse, trade, monitor and make decisions (FCA). Asset managers and investment banks that assumed the Mills Review was a retail story should revisit that assumption. Supervisory interest in model governance, trade surveillance and conflicts management will likely intensify.

Positioning before the next survey

The FCA's survey cadence creates its own dynamic. With 86% adoption now baseline, the firms that stand out in the next iteration will be those that can evidence not just use, but governance maturity. Boards should assume that the regulator will increasingly triangulate self-reported survey data against what it sees in the AI Lab and in supervisory engagement.

The practical implication is that AI governance is no longer a technology committee topic. It is a board accountability question, and the window to get the structure right before testing begins in earnest is closing.