Delaware’s Court of Chancery read Empery Digital’s official minutes side by side with the AI transcripts of the same meetings, and the transcript supplied the motive the minutes left out. The Delaware Supreme Court affirmed on October 8. This issue explains why holding off on AI transcription is the best course for now, why a ban still fails if directors bring personal notetakers, and the six conditions a board must meet if it proceeds anyway.

KPMG and the University of Melbourne found that 66% of employees using AI at work rely on its output without evaluating accuracy, and expert sign-off often fails to fix that when the review happens inside the same chat that produced the answer. Stanford researchers found models change their answers under user pushback 58% of the time, which means an expert can be pulled into the tool’s own reality. This issue explains how to anchor expert review with checkpoints written before the prompt runs and external references the AI cannot influence.

IBM's 2026 survey found 76% of organizations now have a chief AI officer, up from 26% a year ago, which means most of those seats are heading into their first bonus cycle. Federal banking guidance, state insurance disclosures, HHS-OIG, and DOJ prosecutors all reject paying control roles primarily on the results they are supposed to check. This issue explains how to split the bonus 50/50 between deployment and controls, and the four pass-or-fail measures that give examiners real evidence.

Most firms write the Head of AI role as either an adoption mandate or a governance mandate, rarely both, which means the executive holding stop authority doesn't own the upside and every halt becomes one executive overruling another's program. The NAIC's model bulletin and the Fed, OCC, and FDIC's revised guidance both point toward named, senior accountability sitting in one place, the way SOX Section 302 already puts capital allocation and its controls under the same CFO. This issue lays out the eight-question checklist for whether your Head of AI holds the title with real authority or just half of a split mandate.

The NAIC's model bulletin on AI, now adopted by 24 states and DC, tells examiners to request validation and testing documentation, model drift evaluations, and proof that third parties met their obligations. An ISO 42001 certificate proves a vendor manages AI across its business, but it says nothing about the specific model pricing your policies or deciding claims. This issue explains how to split the ask between a management-system certificate and a model-level audit report, and what six questions determine whether what you're holding will satisfy an examiner.

Enterprise supplier assessments now ask companies to name the executive accountable for AI governance and produce evidence of periodic reviews, driven by frameworks like NIST AI RMF and the EU AI Act's Article 25(4). Sales operations can pull existing wording from an RFP library, but it can't confirm whether an AI answer is still true or produce the documentation behind it. This issue explains why the executive who owns AI governance needs a seat inside the renewal cycle, and the six questions every major client relationship should be able to answer with evidence.

EY's review of Fortune 100 filings found 99% disclose how often management briefs the board on cybersecurity, compared with just 9% for AI, even though 48% of the same companies claim AI as a focus of board risk oversight. The gap isn't a missing mandate, it's a missing seat: cyber oversight works because the CISO presents directly, letting directors question the person who knows what actually happened. This issue explains what it takes to give AI the same standing calendar slot, with the executive who owns deployments in the chair.

AI-related securities cases accounted for 73% of the $529 billion in market value lost from disclosure claims in the first half of 2026, even though they made up only 12% of filings, and the ZoomInfo complaint shows exactly how an engagement claim collapses without retention data behind it. Reliance on management is a real defense for directors, but only when it's reasonable given what they knew at the time, and AI disclosures now carry an SEC enforcement template that other risk categories don't. This issue lays out the five questions an audit committee should run against every AI statement before approving the language, not after.

More than half of software vendors report that fewer than one in four customers use the AI features already built into their product, and at most firms the decision to leave those features off was an informal call that renews itself by default every month. Meanwhile the pre-AI baseline needed to measure ROI keeps eroding as organizational changes pile up, making the case for AI investment harder to build every quarter the feature sits dormant. This issue lays out the three questions that let you either turn a feature on with a documented basis or leave it off on purpose, with a named person attached to either answer.

Uber burned through its entire 2026 AI budget by April, four months into the fiscal year, and its own COO has said the company still cannot draw a clear line between that spending and any customer benefit. Most delegation of authority matrices assign a limit to every material expense category except AI, because those frameworks were built for fixed monthly bills, not pay-by-consumption pricing that can grow exponentially within a single reporting period. This issue explains why disclosure and dashboards are not a control, and what an enforced, function-specific spend limit actually requires.

AI questions now appear on applications for general liability, cyber, tech E&O, D&O, EPLI, and fiduciary liability, and at most firms each one gets answered by whoever sits closest to that particular renewal. In January 2026 the Insurance Services Office published endorsements letting insurers strip generative AI from general liability policies, and W.R. Berkley excluded AI absolutely across three of its lines using a definition broad enough to catch predictive models firms have run for years. This issue lays out what each policy actually asks and why naming a single owner is the only way to walk into renewal with a consistent, defensible answer.

A Delaware court unwound Krafton's attempt to oust the Subnautica studio founders after its own CEO's AI chat history surfaced the strategy behind the maneuver, adding 258 days to the earnout period. A federal court in New York has separately ruled that chatbot exchanges carry no attorney-client privilege, meaning the other side in a dispute can compel and read them. This issue explains why AI risk in regulated firms and deal teams is a privilege problem owned jointly by the CCO and General Counsel, not a data security item monitoring tools can solve.

Article 50 of the EU AI Act takes effect on August 2 and requires any AI system interacting with people to disclose that fact from the first exchange. Most compliance teams assume their chatbot vendor's contract language covers this, but Article 3 splits the obligation by role, and the vendor is not automatically the provider for your deployment. This issue walks through the decision tree for determining provider status across customer-facing, HR, and IT chatbots before the deadline.

J.P. Morgan replaced its ISS and Glass Lewis subscriptions in January 2026 with a proprietary AI voting engine, and Kekst CNC's analysis of four major AI models found they backed the activist slate or a split card in 59 to 77% of contested meetings tested -- well above the historical ISS and Glass Lewis management support rate. The models doing that scoring aren't stable either, with re-runs of identical queries changing the vote in 38% of contests. This issue explains what your General Counsel and Head of Investor Relations need to run before your next proxy season, and what the AI is actually being scored against.

Workiva, Diligent, and AuditBoard all made AI-centered product announcements at IIA GAM in March 2026, promising to draft process narratives, generate audit plans, and surface board-ready risk insights. What none of them addressed is whether the data inside your platform is clean enough to make those outputs reliable -- and Workiva's own 2025 survey found only one in three practitioners has data of sufficient quality for AI use. This issue explains what a structured data readiness assessment requires before any AI feature gets turned on.

Most organizations have a governance committee, an acceptable use policy, and a working sense of which AI tools are in use -- but 88% of firms deployed AI before a formal inventory existed, according to McKinsey, which means every approval decision is being made against a picture no one has fully confirmed. Waiting to build the inventory until the Head of AI role is filled trades program speed for a clean handoff, and the deployment acceleration that governance is supposed to enable never arrives. This issue explains who should commission the baseline inventory today and what a complete use case card requires.

The SEC's first AI washing enforcement action against Presto Automation made two distinct charges, and the second one didn't require proving any AI statement was false. It only required showing no process existed to verify the claims before publication, which is exactly the gap that legal sign-off alone leaves open. With 72% of S&P 500 companies now disclosing AI as a material risk, this issue lays out the five-part test for whether your firm has a real disclosure control or just a legal review habit.

Mid-market PE firms are building AI narratives into exit materials to capture valuation premiums, but the claims in a Confidential Information Memorandum travel with the deal as representations that survive closing. The SEC charged two registered investment advisers in 2024 for overstating AI capabilities, and AI-related securities class action filings hit a record pace in the first half of 2025, establishing the enforcement template that is now migrating into R&W insurance claims. This issue explains what documentation a portco must produce before any AI claim goes into sale materials, and why buy-side diligence is increasingly equipped to find the gaps.

Grant Thornton's 2026 AI Impact Survey found that more than half of COOs are concerned about regulatory and compliance failure from agentic AI, while fewer than one in five CIOs and CTOs share that concern. With 83% of S&P 500 companies now disclosing AI as a material risk, a board that receives only the consensus management view has no way to demonstrate active oversight when litigation discovery or a regulatory inquiry arrives. This issue explains why the audit committee is the right standing mechanism to surface that divergence, and the four structural requirements that make it functional.

Courts have already rejected the argument that vendor terms of service shield deploying enterprises from liability -- Air Canada and the Workday and Eightfold AI cases made that clear. A SOC 2 report tells you a vendor controls their operating environment; it says nothing about whether their model produces biased outputs against your specific customer population or drifts from its approved behavior. This issue explains the three independent verification capabilities every regulated enterprise needs to build before the next model update.