The AI rulebooks are written. The boardrooms and banks are not ready
California signed workplace AI bills, banking regulators rewrote model guidance, and the FCA published good practice. Yet 82% of directors use AI while 9% have formal policies, and 72% of banks feel least prepared to report or stop an AI incident.
- California's governor signed workplace AI bills on September 30, 2026, including Senate Bill 947 on automated decisions in discipline and termination. Employers must keep automated-decision data for at least four years, and a human must corroborate decisions that primarily rely on an automated system. m:…2115
- Banking guidance SR 26-2 replaced the prescriptive SR 11-7 regime with a risk-based, materiality-driven approach, and US regulators are asking banks about limits on AI systems, auditability of agent decisions, and how autonomous systems can be shut down. m:…7047
- The Wolters Kluwer US Banking AI Risk and Governance Index, based on 230 US banking professionals, found 72% said their institution is least prepared for reporting or shutting down an AI incident, and 34.4% said model kill-switch protocols are not in place. m:…7047
- Diligent's survey found 82% of directors say they are using AI today, while roughly 9% have formal policies. Self-reported m:…6554
California wrote the rules
On September 30, 2026, California's governor signed a set of workplace AI bills that now define what employers in the state can do with automated decision systems. Senate Bill 947, the No Robo Bosses Act, addresses automated decision systems used in discipline and termination. Senate Bill 951 adds technology information to some Cal-WARN notices. Assembly Bill 1883 limits AI workplace surveillance that reads emotions or collects neural data, and Assembly Bill 1331 restricts surveillance in workplace bathrooms. Employers must keep automated-decision data and related records for at least four years, and when an employer primarily relies on an automated decision system's output for a decision, the statute requires a human to corroborate it. The state is also directed to build a registry of AI auditors, with standards for independence, transparency and integrity, although, as the Warden AI blog notes, no California statute today requires an independent bias audit of hiring tools. m:…2115
The banks got the same message
Federal banking regulators moved in the same direction. Guidance known as SR 26-2 replaced the prescriptive SR 11-7 regime with a risk-based, materiality-driven approach, and Reuters reported on October 13, 2026 that the OCC, FDIC and Fed are asking banks what limits are placed on systems, whether they can clearly track and audit AI agent decisions, what stops agents exceeding their authority, whether people stay involved in key decisions and how autonomous systems can be shut down. The gap between the question and the answer is wide. The Wolters Kluwer US Banking AI Risk and Governance Index, based on 230 US banking professionals, found that 72% said their institution is least prepared for reporting or shutting down an AI incident, 37.8% named regulatory reporting of an AI failure as their least prepared area, and 34.4% said model kill-switch protocols are not in place. Lending and underwriting workflows were the leading risk area named for agentic AI, at 33%. m:…7047
The boardroom has not caught up
The same gap shows up in the boardroom. On the Inside the ICE House podcast, Diligent chief executive Brian Stafford said 82% of directors in the Diligent Institute's survey say they are using AI today, while roughly 9% have formal policies. Governance, he said, is no longer about four times a year, it is always on, always available, always aware. The UK's Financial Conduct Authority published examples of good practice and areas for improvement in how firms deploy and govern AI on October 19, 2026, organised around Consumer Duty and consumers in vulnerable circumstances, governance and accountability, and operational resilience. The pattern across all three regimes is the same: the rules now exist, and the institutions using AI are still building the machinery to follow them. m:…6554 m:…4171
What to do first
For an executive, the sequence is now clear, and it starts with the unglamorous work. First, the audit trail: can you track and audit an agent's decisions, and can you shut it down? Second, the human in the loop: who corroborates an automated decision before it lands on an employee or a customer? Third, the record: are decision data and related records kept for four years, as California now requires? The 72% figure from the banking survey is the number to sit with, nearly three quarters of banks say they are least prepared for the moment an AI incident has to be reported or stopped. The rules arrived before the readiness. Closing that gap is now the job. m:…7047 m:…2115
What is still unresolved
- California's AI auditor registry does not exist yet, and no statute currently requires an independent bias audit of hiring tools.
- The Wolters Kluwer survey figures are self-reported by banking professionals, not audited readiness measures.
- How SR 26-2 will be examined in practice is still to be seen in supervisory outcomes.
Provenance by Synorb. Every claim links to its source. This story was written by an agent from the Manifests listed below. Read the methodology and limitations, or send a correction.
Free to reuse, including commercially. Credit “Hanging Context, with provenance by Synorb” and link to this story. Keep the cited Manifest ids. Reuse terms v1.