Redress Pathways Associate

FCA secures new Leeds office to boost presence outside London. The regulator has committed to have 1,000 roles based outside London by 2030 The FCA has procured a new base in the north of England as part of its commitment to have 1,000 roles based outside London by 2030. The regulator has signed an agreement for a new 36,097 sq ft office at 2 Wellington Place and will move into the space in 2028. "Our Leeds office, currently at 6 Queen Street, has continued to grow since opening in 2022," the FCA said. "There is still some time before the move, and we'll share more about our new office and plans closer to 2028."

Hundreds rush for AI jobs at 'data led' UK financial regulator. Tuesday, 20 Oct 2026 | 11:00 AM MYT "Hundreds" of applicants have thrown their hats in the ring for senior AI jobs at the UK's top financial regulator, according to the new hires' future boss, despite the fact that AI companies can pay higher compensation. Jessica Rusu, the Financial Conduct Authority (FCA)'s chief data information and artificial intelligence (AI) officer, said candidates valued the opportunity to solve "fascinating and interesting problems" at the FCA. "We have two senior leadership roles in the market now and we have had hundreds of really high quality applications," the executive, who spent more than 20 years in the private sector before joining the FCA in 2021, said in an interview. The FCA has been advertising for a head of agentic AI and data science research, who will join the organisation's senior leadership team for a salary of between £140,000 and £190,000 (US$186,000 or RM759,233 and US$253,000 or RM1.03mil) in London, or £126,000 (RM684,841) to £171,000 (RM929,427) if they work from the watchdog's Leeds or Edinburgh offices. Powered by: The regulator is also hiring a head of department, AI and technology security operations within the same salary range. Both roles recently closed for applications. It's part of the FCA's goal to become a global leader in AI regulation, both through its own use of the technology and its oversight of usage across the 35,500 firms it supervises. Rusu told a Bloomberg event in London on Oct 19 that the FCA was building AI tools to help oversee anti-money laundering controls at 60,000 extra firms, as part of a new duty that comes into effect from 2028. It's also using automatic and agentic AI agents to help supervise around 9,000 small wholesale market transaction-reporting firms. The FCA is monitoring the financial services industry's use of AI and encouraging good practices. At the event, Rusu called for firms "to help consumers understand the limits of financial information and advice generated by their models through clear, proportionate and consistent disclaimers." The regulator is also working with firms on cyber defences and the potential risks from AI, as highlighted by the security concerns around Anthropic's new Mythos model earlier this year. "AI is moving faster than any regulatory framework was designed," Rusu said. "That means regulators will increasingly need to make judgments without precedent, while ensuring that trust remains at the heart of financial services." If the FCA had unlimited resources, Rusu said, "we would go after more and more financial crime." The regulator is seeing a rise in AI-enabled fraud, such as deepfake videos encouraging fake investment opportunities, she added. "It really undermines growth." - Bloomberg

FCA: 'Our focus is testing, assurance and monitoring' The UK's Financial Conduct Authority has warned that financial firms are deploying automated customer journeys without sufficient testing, as the regulator called for continuous assurance of increasingly autonomous AI systems and stronger preparations for operational failure. Speaking at Bloomberg's UK Policy Series on October 19, Jessica Rusu, the FCA's chief data, information and intelligence officer, said firms must look beyond whether AI is safe at launch and assess how its behaviour changes in production. "A system that is safe at deployment may behave differently at scale, under stress or as the models beneath it change," Rusu said. "Agentic systems therefore require continuous, conditional assurance. Our focus is testing, assurance, monitoring and surveillance," she declared. The warning puts ongoing testing, monitoring and digital resilience at the centre of the regulator's approach to agentic finance, where systems can increasingly initiate transactions and act on behalf of customers. It also comes alongside the publication of examples of good and poor AI practice, drawn from the FCA's observations of firm testing this year. Rusu identified consumer outcomes, accountability and operational resilience as the three central areas of focus. For QA and quality engineering teams, the implications extend across the lifecycle of an AI service. Initial validation provides only part of the evidence needed to assess systems whose underlying models, operating conditions and interactions can change after release. Rusu said the FCA had already observed weaknesses in how firms develop and deploy AI, including applications without a clearly defined customer need, exaggerated claims about capabilities and automated customer journeys introduced without sufficient testing. The regulator has also seen firms exclude vulnerable consumers by default, raising questions about whether testing adequately covers the people who will use a service and the circumstances in which they may need additional support. "Our focus is testing... directly with the industry to understand how AI systems are being deployed." - Jessica Rusu Examples of stronger practice included a firm developing a live list of vulnerability signals that its chatbot could use to identify customers needing referral to human advisers. Another established an independent ethics panel to scrutinise proposed AI applications before deployment, including questions of ownership, bias and controls. However, Rusu warned that governance investment was failing to keep pace with deployment. "But we have also seen firms retrofit governance after technology has already been deployed or fail to adapt monitoring as systems change," she said. "That is the wrong way around." She called for clear accountability, controls proportionate to risk, human oversight and escalation triggers, with senior managers personally accountable for AI systems. For testing teams, those observations point to a wider assurance challenge: establishing whether the controls surrounding an application remain effective as the application itself evolves. Monitoring that was appropriate at launch may become inadequate following changes to a model, its use or the decisions it supports. Rusu's most direct message on digital resilience concerned preparations for disruption. "Third: stress-test AI systems. Plan for failure. Maintain fallback options. Understand supplier concentrations and dependencies, including reliance on a single model," she said. "A system that fails at scale disrupts the services consumers rely on." Broader strategy. That warning connects AI assurance with the FCA's broader scrutiny of shared technology infrastructure. A service may depend on an external model, cloud platform or data provider, making the resilience of those dependencies relevant to its overall reliability. For QA teams, this points towards scenarios that examine how a complete service behaves when a supplier becomes unavailable or an underlying component behaves unexpectedly. Fallback arrangements and escalation processes need to work under the same conditions that disrupt the primary service. The emphasis builds on the regulator's earlier calls for practical evidence of AI controls. In May, the FCA reopened its AI Input Zone to gather examples covering governance, resilience, deployment, oversight and assurance. Colin Payne, the regulator's head of innovation, summarised that exercise in a short instruction: "Not theory. Evidence." The latest observations give that evidence-gathering exercise a more concrete operational focus, identifying shortcomings in testing and governance while setting out the practices the FCA wants firms to adopt. They also reinforce the connection between AI assurance and third-party resilience highlighted in QA Financial's earlier coverage. In July, FCA director Mark Francis stressed that operational resilience needed to address disruption across commonly used service providers, as well as risks within individual firms. "Operational resilience can't solely be about understanding risks within individual firms," Francis said. "It is also about understanding how disruption at commonly used critical service providers could affect the wider system." Rusu's warning about reliance on a single model brings that concentration concern directly into AI deployment. The stakes are increasing as AI moves from supporting decisions to executing them. Rusu described agents that can search across products, manage payments and investments, and transact with other agents, creating another layer of technology between institutions and customers. "Poorly governed AI can create consumer harm, cyber vulnerabilities and operational failures, potentially at much greater speed and scale," she warned. Rusu also linked AI failures to regulatory incident visibility. "Our rules already require firms to report major incidents, and that must extend to significant AI failures and near-misses." That makes the ability to identify and escalate unexpected AI behaviour an important part of assurance. For testing teams, the practical question is whether monitoring and incident processes can recognise a significant failure or near-miss quickly enough for the firm to respond. The FCA plans to work with industry, other regulators and frontier AI laboratories on a framework for agentic finance, including approaches to agent identity, delegated authority, accountability and execution. Rusu said its response to the Mills Review, published in July, would involve direct engagement with firms on deployment. "Our focus is testing, assurance, monitoring and surveillance: directly with industry to understand how AI systems are being deployed." The regulator is also developing its own use of AI agents and machine-readable data collection APIs, a shift that could put banks' underlying data capabilities under greater scrutiny. Steve Round, president and co-founder of SaaScada, said: "An agentic FCA that can pull data straight from firms through APIs, and gather evidence from every source it can reach, is a warning to every bank still running on legacy systems. Inconsistencies that once went unnoticed will now surface in hours." "Banks have to be able to show how every decision was made, whether a person or an AI agent made it, and that evidence comes from the data," he added. "If the core system can't give a real-time, complete record of what happened and why, banks should expect the FCA to intervene, and at scale," Round concluded. For banks and insurers, the speech sharpens the case for treating AI assurance as an ongoing operational responsibility. The test is whether systems remain controlled as autonomy increases, dependencies change and real-world conditions put their safeguards under pressure. See you next month? October 20, 2026 October 19, 2026 October 16, 2026 October 15, 2026 October 14, 2026 October 13, 2026 October 12, 2026