AI governance, operational resilience, and continuous evidence-ready controls reshape risk work
The gist
Risk management is shifting from periodic policy review to continuous, evidence-backed control execution, with heavier scrutiny on AI, resilience, third parties, and AML.
This week’s developments
AI Governance Shifts from Policy to Audit-Ready Control Testing
California moved AI oversight into auditable compliance this week. AB 1405 now requires AI-audit reports to document scope, objectives, results, evidentiary support, deficiencies, and adherence to internal safety standards, backed by a signed and dated compliance statement. CPPA automated-decision rules add written risk assessments, retention of supporting documentation, and meaningful human review with the ability to challenge or intervene. SB 53 requires public catastrophic-risk testing disclosures and incident reporting to California’s Office of Emergency Services within 15 days, or within 24 hours for imminent risk of death or serious physical injury. AB 2013 adds training-data documentation effective Jan. 1, 2026.
Outside California, Australia proposed AI risk reporting for misaligned behavior and cyber incidents on Nov. 10, New York opened RAISE Act registration with 72-hour critical safety incident reporting, and Alabama said state bank exams will begin using the CSBS AI Supervisory Framework in 2027. The FCA’s scrutiny of AI in AML workflows points the same way: regulators want evidence that AI changes are controlled, monitored, and escalated properly.
For practitioners, the career edge is shifting to people who can produce control evidence on demand. Expect more value for maintaining audit-ready documentation, tracking exceptions continuously, and coordinating incident response across model risk, compliance, operations, and vendors.
How do we build audit-ready AI controls and evidence now?
If you're an individual contributor
- AI work is now judged by evidence, not just intent.
- Get fluent in audit trails, testing artifacts, and exception logs; that’s how you stay indispensable when reviews turn evidentiary.
If you manage a team
- Your team’s value shifts from monitoring models to proving control.
- Coach for documentation discipline, challenge escalation, and human-review rigor so the team can defend AI decisions under scrutiny.
Sources
- The most underrated AI investment: Building the humans in the loop - BusinessToday — Business Today, October 20, 2026
How to train reviewers, document decisions, and escalate issues so AI outputs withstand audit and scrutiny.
- HSCC publishes AI clinical resilience playbook to help healthcare organizations manage patient safety, cybersecurity risks - Industrial Cyber — Industrial Cyber, October 14, 2026
Playbook for documenting, testing, and escalating AI risks in healthcare workflows under regulatory scrutiny.
- Episode 25: Human in the Loop Is Not a Strategy — Finding 12 Minutes Podcast, September 22, 2026
Learn how to match human review, escalation, and documentation to AI task risk and complexity.
If you lead the organization
- AI governance is becoming an audit and incident-response operating model.
- Fund control testing, retention, and cross-functional escalation now; regulators want proof your AI is monitored, not just approved.
Sources
- The AI governance moment: Why boards must treat AI risk as an enterprise risk — Fortune India, September 21, 2026
Learn why boards must fund continuous AI control testing, monitoring, and escalation to satisfy regulators and manage enterprise risk.
- The Missing Layer in Enterprise AI: A Governance Architecture for LLMs and Machine Learning | HackerNoon — HackerNoon, October 23, 2026
Shows how to add automated testing, logging, and escalation so AI governance proves ongoing control, not just approval.
Operational Resilience Moves From Control Testing to Service Survival
UK regulators tightened operational-resilience expectations across cyber, incident reporting, and market infrastructure. On 18 March 2026, the FCA, PRA, and Bank of England published a unified incident-reporting regime taking effect on 18 March 2027, requiring initial notifications as soon as practicable, material-change updates, and reporting of significant third-party incidents through one common approach. The NCSC also launched its Cyber Adversary Simulation scheme, using NCSC-assured providers to run bespoke simulations from reconnaissance through remediation. In parallel, draft cyber-resilience rules require firms and financial market infrastructures to identify critical services, set impact tolerances, map dependencies, and test severe-but-plausible scenarios; trading venues must be able to handle at least twice their highest message rate in the prior five years, resume trading within or close to two hours, and keep data loss close to zero. Malaysia’s planned data-centre law extends the same scrutiny to infrastructure itself.
The shift is clear: regulators are moving from checking controls to testing whether essential services survive disruption. The FCA is also treating poorly governed generative and agentic AI as an operational and systemic risk, while using its own AI market first responder to oversee about 9,000 small wholesale transaction-reporting firms. For risk teams, the job is becoming dependency mapping, scenario design, incident triage, and third-party oversight under stress.
How should we prove critical services survive real disruptions?
If you're an individual contributor
- Control testing is fading; your value is in service survival under stress.
- Build skill in dependency mapping, incident triage, and scenario testing—those are the tasks that will make you indispensable.
If you manage a team
- Your team is moving from checklist work to resilience under real disruption.
- Coach for judgment, escalation, and third-party oversight; reallocate time from routine testing to severe-scenario prep.
Sources
- How Enterprise Teams Can Build an Incident Response Process That Holds Up Under Pressure | HackerNoon — HackerNoon, November 13, 2026
Defines severity, ownership, escalation, and postmortems to keep incident teams coordinated under stress.
- Cyber Resilience Is A Business Operating Model, Not An IT Recovery Plan — Forbes, November 4, 2026
How to govern cyber resilience around critical services, escalation, and severe-scenario recovery planning.
If you lead the organization
- Your operating model must prove services survive, not just controls pass.
- Invest in resilience testing, common incident reporting, and AI/third-party governance before regulators force the redesign.
Sources
- Operational Resilience Amid Geopolitical Tension: Key Shifts | Forvis Mazars US — Forvis Mazars US, October 6, 2026
Learn how leaders must test critical services, map dependencies, and strengthen governance before disruptions escalate.
Risk Workflows Are Moving Into Centralized, AI-Enabled Platforms
On November 12, 2026, Charlotte said it will replace legacy web forms, SharePoint, and department-specific workflows with DXC Technology’s AI-powered Assure Risk Management platform, consolidating incident reporting, health and safety management, document processing, and real-time reporting in one environment tied to its claims system. Wisconsin’s statewide GRC modernization points in the same direction: public-sector risk teams are moving away from fragmented manual processes and into centralized platforms.
The shift is not just administrative. The cited research shows automation being applied to exposure tracking, document extraction, control testing, sanctions and fraud screening, anomaly detection, and alert prioritization, with agentic AI extending into investigation support, remediation drafting, and escalation while keeping audit logs and requiring human approval for high-impact decisions. McKinsey estimates referenced in the coverage put the upside at up to 70% of manual risk tasks eliminated and 30–60% efficiency gains after human validation.
For practitioners, the implication is direct: risk work is becoming a platform discipline, not a spreadsheet-and-email function. Teams that can standardize workflows, define approval thresholds, and manage exceptions will move faster; those that cannot will spend more time reconciling data than managing risk.
How should we redesign roles as risk workflows move to AI platforms?
If you're an individual contributor
- Manual risk work is shrinking; judgment and AI oversight are the edge.
- Learn to validate AI outputs, handle exceptions, and document decisions—those skills will keep you valuable as forms and reporting automate.
Sources
- 1,368 AI Failures and What They Teach About Evals — Vishal Srivastava & Tanmay Sah, New York Life — AI Engineer, October 22, 2026
Learn a build-break-fix method to find AI failure zones, add safeguards, and route exceptions by risk.
If you manage a team
- Your team's value shifts from processing cases to coaching decisions.
- Rebalance time toward workflow design, QA, and exception handling; coach staff to supervise AI and escalate high-risk calls fast.
Sources
- AI Can Do the Work. But Who Owns the Workflow? — The AI Maker, September 29, 2026
Shows how to map workflow steps, assign AI responsibilities, and preserve accountability in support, sales, and finance.
If you lead the organization
- Your operating model is overdue for a platform-led risk redesign.
- Invest in centralized workflows, approval thresholds, and AI governance now—or keep paying for fragmented teams to reconcile data.
Sources
- Enterprise AI risk management: Why approval determines who scales - Express Computer — Express Computer, October 19, 2026
Shows why centralized authorization, governance, and monitoring determine whether enterprise AI reaches production safely.
- The Next AI Challenge Is Not Intelligence. It Is Control — Global Banking & Finance Review, October 27, 2026
How to control AI agents with approvals, limits, and oversight before they act inside business workflows.
- AI risk management needs to move beyond individual models — Consultancy.uk, October 2, 2026
How firms need consolidated AI inventories, continuous monitoring, and governance beyond single-model approvals.
Third-Party Risk Is Moving to Continuous Nth-Party Mapping
Ireland’s Central Bank has proposed expanded third-party risk guidance that would force firms to look beyond direct vendors and assess indirect concentration in sub-outsourcing chains, including provider lock-in, geographic concentration, substitutability constraints, and contagion from shared underlying providers. It also tightens subcontractor oversight with prior approval for sub-outsourcing, conditions and monitoring for those arrangements, and registers for critical or important services that capture the full subcontractor chain. In parallel, DORA is pushing firms toward continuously updated registers of ICT providers and their subcontractors rather than static procurement inventories.
The practical shift is clear: regulators are operationalizing continuous visibility into nth-party dependencies and systemic concentration risk, especially for cloud and other critical ICT services. For risk teams, periodic vendor reviews are no longer enough on their own. Expect more frequent dependency-data updates, stronger governance over subcontractors, and better detection of where multiple “separate” vendors rely on the same infrastructure. This raises the bar for business-impact analysis, impact tolerances, and exit planning, and it makes third-party oversight a live operating discipline rather than an annual compliance exercise.
How should we map and monitor nth-party concentration risks continuously?
If you're an individual contributor
- Periodic vendor checks are fading; nth-party mapping is now your edge.
- Learn to trace subcontractor chains and spot shared infrastructure risk; that judgment will make you harder to replace.
If you manage a team
- Periodic vendor checks are fading; nth-party mapping is now your edge.
- Learn to trace subcontractor chains and spot shared infrastructure risk; that judgment will make you harder to replace.
Sources
- A four-week plan to tackle vendor concentration risk - Help Net Security — Help Net Security, September 29, 2026
Learn a four-week method to uncover subcontractor choke points, shared infrastructure risk, and exit-ready contract safeguards.
If you lead the organization
- Periodic vendor checks are fading; nth-party mapping is now your edge.
- Learn to trace subcontractor chains and spot shared infrastructure risk; that judgment will make you harder to replace.
Continuous, Evidence-Ready Controls Replace Siloed Compliance Reviews
Poland and Switzerland have put fixed dates on a tougher cross-border control regime. Poland will apply the EU Anti–Money Laundering Regulation directly from 10 July 2027, lowering the beneficial-ownership threshold to 25% or more, widening the set of obliged entities, and requiring sanctions-evasion risk to be embedded in risk assessments and procedures. Switzerland’s revised FINMA AML ordinance takes effect on 1 January 2027 and makes sanctions compliance an explicit organizational duty under the Embargo Act, backed by documented risk analysis, board-approved directives, a qualified responsible function, trained staff, and screening before transactions or relationships proceed.
At the same time, crypto firms are being pushed toward banking-grade controls, including enhanced KYC, transaction monitoring, suspicious-activity reporting, OFAC screening, and reserve or capital safeguards. U.S. supply-chain rules are also reshaping Israeli defense procurement, with FMF off-shore procurement allowances falling from 25% in FY2019 to 0% in FY2028 and co-production shifting to U.S. facilities such as Raytheon-Rafael’s R2S plant in Camden, Arkansas.
For risk, compliance, and procurement teams, the job is moving from periodic review to continuous control orchestration. Your edge will come from building one auditable operating record that ties ownership, sanctions exposure, transaction behavior, and supplier eligibility together across jurisdictions.
How do we redesign controls for continuous, auditable compliance?
If you're an individual contributor
- Periodic checks are fading; evidence-ready control work is the new edge.
- Build fluency in ownership, sanctions, and transaction screening—your value shifts to spotting issues before they become audit findings.
If you manage a team
- Your team must move from review cycles to always-on control execution.
- Coach for documented judgment, escalation discipline, and cross-border consistency; manual checklist work won’t scale.
If you lead the organization
- Your operating model is being judged on continuous, auditable control coverage.
- Invest in one control record across compliance, procurement, and sanctions—or you’ll keep funding siloed reviews that miss risk.