Securities Filings Hit 118 in H1 2026 as D&O Pricing Sits at Cycle Bottom

NERA economists Ivelina Velikova and Svetlana Starykh counted 118 federal securities class actions in the first half of 2026, an annualized pace of 236 that would top 2025's 205 filings and mark the highest full-year total since 2020 (NERA Economic Consulting, July 21, 2026). Eighteen of those filings targeted AI-related disclosures, already more than the 17 filed in all of 2025, while the directors-and-officers market pricing that book of risk is running flat to down.

A Filing Count Reports Say Is the Highest Since 2020

NERA's own breakdown splits the 118 first-half filings into 67 in the first quarter and 51 in the second, with 112 of the 118 involving standard Rule 10b-5, Section 11, or Section 12 allegations. Cornerstone Research, working from a partially overlapping but not identical dataset, counted 121 core federal filings over the same window, a 30% jump from the second half of 2025, split 69 in Q1 and 52 in Q2 (Cornerstone Research, July 29, 2026). The two counts differ by a few cases because the firms apply different inclusion criteria to borderline filings, a routine methodological gap between the field's two main trackers rather than a data error, and both point the same direction: filings are accelerating, not drifting.

The annualized 236 figure, if H1's pace holds through December, would sit three cases above the 233 filed in 2023, the prior post-2020 peak, and 15% above 2025's 205 (NERA, July 2026). The D&O Diary's summary of the same NERA data adds texture on where the growth and the shrinkage are each concentrated: pump-and-dump allegations went from no more than two filings a year across 2022 through 2025 to 11 in H1 2026 alone, while crypto-related filings fell to 2 from 14 across all of 2025 and SPAC-related filings fell to 1 from 5. Foreign issuers, meanwhile, made up 20.5% of H1 filings (23 cases) against 13.8% in 2025, still under their 29.8% share of U.S.-listed companies but rising fast, a pattern Cornerstone links to a wave of "pump-and-dump" claims against non-U.S. issuers, on pace for 46 filings in 2026 against roughly half that in 2025.

The AI-Disclosure Category Outgrowing Every Established Loss Driver

The AI-litigation trend is the sharpest line in the data. NERA counted 18 AI-related securities filings in H1 2026, already exceeding the 17 filed across all of 2025 (NERA, July 2026); Cornerstone's own AI-specific count runs slightly lower at 15, still on pace to roughly double 2025's total. What separates the two firms' figures more than the count is the dollar exposure each attaches to the category. Cornerstone Research's Disclosure Dollar Loss Index, a measure of the market-cap decline attributable to alleged misrepresentations at the moment they became public, hit $529 billion in H1 2026, up 77% from the second half of 2025, and AI-related cases alone accounted for $385 billion of that, or 73% of the total, despite representing only 13% of core filings (Cornerstone Research, July 29, 2026). Eighty percent of the period's mega-MDL filings, the largest and most complex multidistrict litigation dockets, were AI-related. Cornerstone senior vice president Alexander "Sasha" Aganin attributed the sector shift directly to the technology: "Primarily fueled by AI-related concerns, filings targeting the technology sector increased by 15 filings, from nine to 24" (Alexander Aganin, Cornerstone Research, July 29, 2026). Stanford law professor and former SEC commissioner Joseph Grundfest framed the asymmetry in the same release: "AI-related cases represented a modest share of total filings but an outsized share of alleged investor losses in the first half of 2026" (Joseph Grundfest, Cornerstone Research, July 29, 2026).

That asymmetry, a small share of case count carrying a large share of alleged loss, is the actuarial signal underneath the headline count. A frequency trend built off the case-count trajectory alone understates where the severity is actually concentrating, because AI-disclosure suits are disproportionately landing against larger-capitalization technology issuers where a single adverse disclosure moves a bigger market-cap number than the median filing in the book. A separate market analysis found 13 AI-related federal securities filings in the same six-month window against 14 for all of 2025 by its own count, a third figure that lands between NERA's and Cornerstone's and underscores that every tracker agrees on direction even where the exact count varies (Insurance Business, 2026).

Settlement Frequency and Severity Behind the Filing Count

The resolution side of the docket confirms the same frequency-severity split. Of 105 federal securities cases resolved in H1 2026, NERA counted 65 dismissals and 40 settlements, putting dismissals ahead of settlements at 62% to 38% (NERA, July 2026). Cornerstone Research's parallel settlements tracker counted 39 settlements totaling $2.2 billion in the period, an aggregate that already represents roughly three-quarters of the inflation-adjusted full-year 2025 settlement total; if the second half mirrors the first, 2026 would produce the highest annual settlement dollar volume since 2020 (Cornerstone Research, cited in The D&O Diary, July 2026). NERA's own average settlement value for H1 2026 came in at $54 million, a 32% jump from 2025's inflation-adjusted average of $41 million, with a median of $18 million (NERA, July 2026). Four mega-settlements exceeding $100 million each accounted for roughly 10% of all H1 2026 settlements by count but a disproportionate share of the dollar total, the same concentration pattern the filing data shows on the AI-loss side.

That combination, higher average settlement value against a still-elevated dismissal rate, is consistent with a docket where the marginal case surviving to settlement is a stronger one, plaintiffs' firms are being more selective about which AI-disclosure claims they pursue past a motion to dismiss, and the ones that do survive carry larger alleged losses because of who is being sued. A frequency-down, severity-up settlement pattern is exactly the shape D&O actuaries already track in the broader liability book; what is new in mid-2026 is a distinct sub-category, AI disclosure, generating that pattern inside a single accident-year cohort at a pace the historical loss triangle has no comparable rows for.

A D&O Market Still Pricing Off the Old Trend

The litigation data landed the same week Marsh published its Q2 2026 Global Insurance Market Index, and the pricing signal it shows for financial and professional lines does not yet reflect the AI-litigation acceleration. Marsh's global composite for financial and professional lines, the category that includes D&O, declined 3% in the second quarter, continuing a run the firm's own release frames as the eighth consecutive quarter of rate reductions across commercial lines broadly, driven by abundant capacity, strong insurer profitability, and favorable reinsurance terms (Marsh, July 22, 2026). U.S. D&O pricing specifically ticked up 1% in the second quarter after a 3% decline in the first, the first positive print for the line in more than two years and a possible early signal that underwriters are starting to price the litigation trend into renewals, though a single quarter of modest firming does not undo four years of cumulative reductions. Marsh's John Donnelly, president of global placement, described the broader posture as one where "insurers are seeking to differentiate through broader coverage, expanded terms, and lower deductibles" even as underwriting discipline sharpens in select lines (John Donnelly, Marsh, July 22, 2026), a description that fits a market defending share on non-price terms rather than one pulling back capacity.

That flat-to-slightly-firming D&O rate sits against a premium base that has already contracted sharply. actuary.info's coverage of AM Best's July 2026 special report put the 2025 monoline D&O direct loss ratio at 54.5, up from 49.0 a year earlier, on direct premium down to roughly $10 billion from nearly $15 billion in 2021, a one-third contraction in four years, while reserves for accident years 2023 and 2024 had already proved inadequate. TransRe's own 2026 market analysis frames the line's expected breakeven around a 70-point ultimate loss ratio after typical acquisition and internal expense loads, which puts 54.5 comfortably under that line on paper, a comparison that only holds if the underlying accident-year reserves are adequate (TransRe, April 2026). AM Best's own report already said they were not.

Where the Margin Gets Tested First: Excess and Side-A Layers

The tower structure of a typical D&O program means the AI-litigation trend will not hit every layer evenly, and the layers most exposed to a new severity category are also the ones that have captured the least pricing since the last hard market. TransRe's data shows mega and large-cap D&O excess-layer pricing sitting 13% to 17% below its 2013 peak despite four years of exposure growth from rising market capitalizations, with low and mid-excess layers carrying the highest sensitivity to severity increases yet the least price appreciation of any tower segment. Brokers describe the phenomenon inside those middle layers as the "burn layer," attachment points that historically sat well above where claims activity reached but are now increasingly breached as severity migrates upward through the tower, with several markets openly calling those layers underpriced relative to current claims trend (Insurance Business, 2026). Side-A difference-in-conditions coverage, the layer that protects individual directors and officers when the corporate balance sheet cannot or will not indemnify them, is one of the specific positions market commentary flags as having limited headroom for further rate cuts given how compressed its margins already are, even as broader excess capacity remains abundant absent a market shock.

A cumulative 44% rise in legal-services fees since 2020 compounds the tower problem independent of the AI-litigation trend, because higher defense costs erode the working layer faster and push more of a given claim's ultimate cost into the excess layers above it (TransRe, April 2026). Layer that structural cost inflation on top of a new claims category concentrated in large-capitalization technology names, exactly the risk profile that tends to buy the deepest towers, and the excess and Side-A segments face two compounding pressures at once: a severity trend the primary layer's pricing does not reflect, and a cost-of-defense inflation rate that was already eating into margin before the AI cases showed up in the docket.

The Reserving Question for Accident Years 2025 and 2026

For a D&O reserving actuary, the practical question the mid-year data raises is not whether AI-disclosure litigation is real, the filing counts settle that, but whether the loss development pattern built from pre-2025 accident years still applies to the vintages now accumulating this new category of claim. A loss triangle populated mostly by pre-2024 filings has no rows for a claims category that Cornerstone Research says did not exist in meaningful volume before 2023 and that NERA's own count shows already exceeding a full prior year's total halfway through the current one. Applying an unadjusted historical development factor to accident year 2025 or 2026 risks understating ultimate losses specifically in the segment of the book where filings are growing fastest and alleged losses are concentrating most heavily.

The practical adjustment is not a wholesale rewrite of the D&O loss triangle but a segmentation of it: carriers with enough volume to isolate technology-sector, large-capitalization exposures from the broader D&O book can track that segment's own emerging development pattern separately rather than blending it into an aggregate triangle where a still-small absolute claim count gets diluted by the much larger base of non-AI filings. For carriers without that volume, the more conservative move is applying an explicit loading to accident years 2025 and 2026 pending two or three more years of maturity on the AI-disclosure cohort specifically, rather than assuming the existing triangle's tail factors already capture a claims driver that is less than two years old. Given that AM Best has already flagged reserve inadequacy in the 2023 and 2024 accident years on the broader book, before the AI-litigation surge fully showed up in the docket, the case for an explicit loading on the newer vintages is stronger than it would be in an otherwise stable claims environment.

What the Reinsurance Read Implies

Financial-lines reinsurers price off the same primary data these filing and settlement figures describe, and a rate environment that is flat to down while the underlying claims trend accelerates is a mismatch reinsurers are positioned to see before many primary underwriters adjust their own pricing. Favorable reinsurance terms are among the factors Marsh itself cites as supporting the eighth consecutive quarter of commercial rate declines, which means the softening at the primary level is being reinforced, not offset, by treaty terms behind it. That dynamic can persist for several renewal cycles even as the underlying loss cost deteriorates, because reinsurance capacity responds to aggregate market conditions and available capital more than to a single emerging litigation category, until enough ceded loss experience from the newer accident years works its way back through treaty results to move pricing. A primary D&O actuary relying on soft reinsurance terms as evidence that aggregate risk is stable should treat that read with more caution than usual this cycle, given how concentrated the AI-litigation dollar exposure is in a small number of large-capitalization names that a handful of adverse outcomes could move materially.

The Renewal-Season Read

The mid-year data leaves D&O underwriters and their actuaries with a specific gap to reconcile: a filing count on pace for the highest annual total since 2020, a settlement-dollar trajectory on pace for the highest since the same year, and a new loss category already carrying 73% of the year's alleged-loss dollar exposure on 13% of filings, against a pricing environment that only turned modestly positive in the U.S. after two years of cuts and remains negative globally. The next data points that will show whether the market is pricing this correctly are the third-quarter 2026 Marsh index, due in October, and Cornerstone Research's full-year 2026 filings report, typically released in January, which will show whether the H1 pace held through year-end or whether the second half's filing count decelerated the way 2023's initial surge eventually did. Until then, the loss triangle for AI-disclosure claims specifically has fewer than two full accident years of data behind it, and a D&O book renewing on a rate that has moved only 1 point in the direction the claims trend has moved 73% of a $529 billion loss index is a book betting the trend does not continue.