Connecticut Printed the Dates. IBM Asked What Boards Did Next.

Picture the meeting where an AI incident was reported to the board. A slide, three questions, a nod, the next agenda item. Everyone in the room felt responsible. Nobody left with a task.

IBM tried to measure how often that happens. In a Censuswide survey of 3,221 executives at companies with 1,000 or more employees in eight countries across Europe and the Middle East, 80% said an AI-related incident had reached their board or C-suite in the past year. Fewer than half of those organizations, 40%, reported a change to AI governance policy in response. When deployment outruns governance, 56% keep going within agreed risk limits while governance catches up. A third wait. The survey did not cover the United States, and I would not assume a U.S. board looks different.

Meanwhile, the calendar is being written in statehouses. Connecticut's CART Act, Public Act No. 26-15, began phasing in on October 1. The first provisions require disclosures for AI subscriptions and add whistleblower protections for people involved in developing or deploying large frontier AI models. Employers that use automated employment decision-making must have a framework in place by October 1, 2027, to tell applicants and employees when and where they are dealing with AI. Chatbot rules follow on January 1, 2027. Sen. James Maroney, the law's lead architect, told a press conference: "This is not a finish; this is not a ceiling. This is the floor."

Both stories carry the same lesson. The dates are printed. What is not printed is who answers for the decision before a date arrives.

IBM's data holds one more number worth sitting with. Ninety-eight percent of respondents say responsibility for AI risk is assigned at board level. Twenty-six percent have a dedicated board committee for it. A committee is not the only way to oversee AI, but it is one way to tell whether anyone owns the question.

You can test your own board in one meeting. Ask three things. Does an AI incident log exist, and who, by name, reviews it? Does that person or committee have the authority to change policy? And does your business operate in, hire in, or sell AI subscriptions to residents of Connecticut?

If the answer to the first question is "we do not have one," that is the first finding.

Read the full research behind this story, plus this week's senior housing and insurance coverage flagships, in the LITT Briefing on Substack.

If you want to put an AI governance decision on paper before the next incident forces one, the ninety-day playbook is in my book, Senior Housing: The AI Gap: From Governance Theater to Deployment in 90 Days, available on Amazon. The compliance checklists and implementation tools are at youraigap.com.

← Back to Governance