Editor’s Note
The most interesting AI developments this week were not really about models getting smarter.
Independent insurance agencies are adopting AI much faster than they were two years ago. Insurance regulators are moving AI tools into the hands of people reviewing forms. Outside the industry, insurers are beginning to appear in conversations about how companies should govern their own AI use. And some of the largest AI companies have now voluntarily agreed to a governance structure built around internal controls, independent review, and board oversight.
Different parts of the market are arriving at a similar place.
AI adoption is getting easier. Deciding how it should operate, who checks it, and who owns the result is becoming the harder problem.
– James W. Moore, Editor-in-Chief
AI Use Among Independent Agencies Jumps From 15% to 46%
The Big “I” and Future One released their 2026 Agency Universe Study with one number that deserves attention: 46% of independent agencies now report using artificial intelligence, up from 15% in 2024.
The most common reported uses among AI adopters include marketing content generation, coverage-form analysis, and contract review.
The same study shows how unfinished that adoption remains. Sixty percent of agencies identified lack of knowledge about AI capabilities as a barrier, while 48% cited security and privacy concerns. Keeping up with AI was itself named one of the industry’s leading challenges.
Meanwhile, three out of four agencies reported revenue growth between 2024 and 2025, average staffing increased from 8.2 to 9.9 employees, and about one-third of agencies increased headcount.
Why it matters: The interesting comparison is not AI versus people. Agencies added AI use and employees at the same time.
Nor does the study establish that AI caused the revenue growth. What it does show is an independent agency channel adopting the technology quickly while still trying to determine where it belongs.
Going from 15% to 46% in two years is no longer early experimentation by a handful of technology-forward agencies. The governance problem increasingly moves downstream as well. Agencies reviewing contracts and coverage forms with AI need to know what information can be entered, what output requires verification, and where the person using the tool remains responsible for the conclusion.
NAIC Puts CLARA in Regulators’ Hands
The NAIC Insurance Summit this week included hands-on workshops for state insurance regulators using CLARA, the Compliance Language Assistant for Regulatory Analysis.
CLARA is designed to assist regulators reviewing insurance forms. In the workshop, reviewers define a regulatory rule, provide examples of compliant and non-compliant language, and test filing language against individual rules or rule sets. The system returns a pass/fail result with supporting rationale.
The NAIC describes the process as reviewer-driven, with the regulator remaining responsible for the decision.
That distinction is important. CLARA is not merely summarizing a filing for someone to read faster. It is beginning to participate in the compliance-review workflow itself.
Why it matters: Shared regulatory infrastructure can spread capability faster than statutes or individual departments can build it independently.
A regulator in one state does not surrender authority because several states use the same technology. But when regulators can define rules, test language, share experience, and improve a common platform, the mechanics of supervision begin to converge even when the legal authority does not.
The questions now become operational ones. How consistently are rules encoded? How are disagreements with CLARA handled? What gets preserved in the review record? And what happens when one regulator develops a particularly useful rule set?
The technology matters. The ability to distribute regulatory capability may matter more.
Insurance Is Becoming Part of the AI Accountability System
A discussion at AdExchanger’s Programmatic IO conference offered an interesting glimpse at where insurance may fit into AI governance outside the insurance industry.
Betty Louie, partner and general counsel at The Brandtech Group, described insurance underwriters as an emerging participant in the AI discussion. Companies seeking coverage may increasingly face insurer questions about their AI policies, human oversight, controls, and auditability.
The discussion also emphasized the ability to reconstruct what happened when an AI-enabled process fails: which model was operating, what information it used, and whether the resulting error came from a person or a machine.
None of that makes insurers AI regulators. It does give them another mechanism for influencing behavior.
Why it matters: Insurance has always converted uncertainty into underwriting questions.
If insurers begin asking companies how AI systems are governed before providing coverage, pricing a risk, or determining terms, AI governance stops being solely a compliance exercise.
Controls can become underwriting evidence.
That could include documentation, model inventories, human-review requirements, audit trails, incident procedures, third-party model governance, and the ability to reconstruct an AI-supported decision after something goes wrong.
The interesting part is not whether an insurer can create another questionnaire. It is whether better governance eventually produces meaningfully different coverage, capacity, or pricing.
That is when accountability becomes an insurance mechanism rather than another box to check.
AI Companies Agree to Four Layers of Oversight
President Donald Trump and executives from several major AI companies signed a voluntary AI accord this week outlining four layers of controls for companies developing frontier models.
The framework calls for internal controls monitoring model capabilities and risks, an internal team responsible for making sure those controls operate as intended, independent external evaluation, and an independent board committee receiving reports and overseeing remediation.
The document was signed by executives representing Google, Anthropic, Meta, OpenAI, X, and Nvidia. It is voluntary rather than a binding regulatory requirement, and the accord itself says the measures could eventually be codified into law or regulation.
That makes the document less important as regulation than as an emerging description of what responsible organizational control might look like.
Why it matters: The structure should look familiar to insurers.
Operational controls. Internal oversight. Independent review. Board accountability.
Those are not uniquely AI concepts. They are familiar ways of governing consequential systems and risks.
For insurers, the more interesting question may be what happens if structures like these gradually become evidence of reasonable AI governance. An underwriter evaluating an organization’s AI exposure could eventually care less about whether it has an impressive AI policy and more about whether responsibility actually travels from the operating system to independent review and ultimately to the board.
Voluntary standards do not automatically become insurance standards.
But once a control becomes recognizable, measurable, and reasonably common, someone eventually starts asking whether it was there before the loss.
Sources
- Independent Insurance Agents & Brokers of America / Future One, “Big ‘I’ and Future One Release 2026 Agency Universe Study”
- NAIC, “CLARA in Action: Practical AI for Form Review,” Insurance Summit 2026
- AdExchanger, “Programmatic IO: Insurers Are Building Ad Tech’s AI Accountability Layer”
- Reuters, “Trump releases AI accord with tech executives”
- Forbes, “Here’s The White House’s 308-Word AI ‘Accord’ In Full”
AI Disclaimer: This content was created with assistance from artificial intelligence technology. While content is based on factual information from the source material, readers should verify all details directly with the respective sources before making business decisions.
