Jessica Rusu, FCA Chief Data, Information and Intelligence Officer, says the regulator wants to be at the forefront of AI adoption.
The Financial Conduct Authority (FCA) has outlined how it plans to regulate firms’ use of AI and revealed its own plans for the technology.
At the Bloomberg UK Policy Series on 19 October, Jessica Rusu, FCA Chief Data, Information and Intelligence Officer, said the financial watchdog is leveraging AI to increase operational efficiency and train staff.
“AI can gather evidence, surface intelligence from disparate sources, and flag early warning signs so we can act at pace,” said Rusu.
“Part of our work to become a smarter regulator means developing our future as an agentic regulator, using the same technologies reshaping our markets to make ourselves sharper, faster, and more effective.”
Rusu made sure to address those in attendance as “humans” at the start of her speech before listing some of the ways the FCA is employing the technology.
She said the regulator is building AI tools to oversee 60,000 anti-money laundering (AML) firms that will come under its remit in 2028, as well as an automated system to monitor around 9,000 smaller wholesale market transaction reporting firms, allowing supervisors to focus on more complex cases.
The FCA is also developing machine-readable data collection APIs to complement its Handbook API, launched in August, which allows firms to access its rules in a format that can be processed by their own AI tools.
Rusu added that the regulator’s AI Lab and Sandbox would play an important role in testing system-wide scenarios and developing new approaches to supervision.
Not downplaying the risks of agentic AI
Rusu said firms are adopting AI at speed, with a joint FCA and Bank of England survey due to be published next week, showing that 86% of firms now use the technology.
Payment Expert has reported on how AI is helping financial services improve fraud detection and prevention in the past. By quickly identifying suspicious patterns, it is helping firms respond to threats as faster payments make speed more critical than ever.
However, while agentic AI could take these capabilities further by allowing systems to act and initiate transactions themselves, it also opens the door to new risks.
Speaking to Payment Expert earlier this year, Thetaray’s VP of Financial Crime and Compliance AI, Garima Chaudhary, warned that agentic payments challenge the assumptions behind traditional AML systems, which rely on human behaviour to identify suspicious activity.
As AI agents begin initiating payments, patterns change, making it harder for existing systems to distinguish legitimate activity from potential fraud.
Rusu acknowledged similar risks in her speech, warning that “poorly governed AI” can lead to consumer harm, cyber vulnerabilities and operational failures.
“That is why firms need to invest now in cyber and operational resilience. Our rules already require firms to report major incidents, and that must extend to significant AI failures and near-misses,” she added.
“As the AI Minister has recognised, near-miss reporting for agentic AI is becoming a national security priority. It is a regulatory priority too.”
She stressed that testing mustn’t stop once an AI system is deployed, stating that agentic AI may behave differently as it scales or operates under pressure.
The FCA plans to work with financial services firms, other regulators and AI developers to establish a framework for agentic finance. This will cover agent identity, delegated authority, accountability and the execution of transactions.
FCA sets out good and poor practice
As part of the address, Rusu noted the FCA has published examples of good and poor practice in firms’ use of AI, stressing the importance of consumer protection, accountability and operational resilience.
The guidance makes use of the regulator’s observations from firm testing this year and aims to help financial services businesses deploy the technology responsibly.
On consumer protection, the FCA said firms should use AI to improve outcomes over automating existing processes. Examples of good practice include designing more inclusive products, providing clear consumer choices and ensuring customers can speak to a human adviser when needed.
However, the regulator also identified firms deploying AI without a clearly defined purpose, excluding vulnerable consumers by default and overstating their systems’ capabilities.
The FCA also emphasised that firms must establish clear accountability, effective governance and human oversight before deploying AI.
Operational resilience was picked out as a further priority, with firms urged to stress-test systems, ensure fallback arrangements and understand their reliance on third-party suppliers.
The regulator said its guidance would evolve at the same time as the technology, adding that well-deployed AI could benefit consumers through more efficient operations, competitive pricing and faster services.
“I began by saying it was good to be here among humans. While that was light-hearted, there is a serious point behind the work we must do together,” concluded Rusu.
“AI is moving faster than any regulatory framework was designed. That means regulators will increasingly need to make judgements without precedent, while ensuring that trust remains at the heart of financial services.”
