← All companies

AIGAmerican International Group, Inc.

Industry Insurance Updated 24 Sep 2026

The idea

The only P&C name in the cluster at or below book (P/B 0.97). The base case is AIG keeps trading at book while earning 11-12% on it; the Corebridge exit is done and underwriting has started growing again.

Events to watch

  • Q3 results ~2026-11-03
  • buyback pace ($2.6bn authority left) against the premium trend
  • Item 5.02 8-Ks after two senior departures in fifteen days
How we got the target

Scorecard model table: base ~$93 ('does not require a re-rating'). The scorecard's summary line says ~$92. Time frame: 2028 model line.

Target: from the research notes. Prices: market data get_latest_stock_price, close of 2026-09-22 (fetched 2026-09-23)

What would prove it wrong

8 warning signs
  1. A third senior departure at Executive Chair / segment-CEO level before the Q3 2026 call, or a General Insurance CEO hired from outside.

    Details

    Two in fifteen days is a pattern; three is a verdict.

  2. U.S. excess casualty strengthening exceeds the workers' compensation release in any quarter.

    Details

    Q2 2026 was $74m adverse against $177m favourable. The favourable line is a book AIG has been running off for a decade; the adverse one is accident years 2016 and 2023, which are not finished.

  3. The accident-year combined ratio as adjusted crosses 90 while net premiums written still grow — i.e. the growth is being bought.

    Details

    It was 88.1 in Q2 2026 and has been 87.7-88.3 for three years.

  4. The buyback returns above operating cash flow while the euro notes are outstanding.

    Details

    AIG has just refinanced EUR 821m of 1.875% paper with 4.250% and 4.750% notes and taken on ~EUR 300m more; debt is 17.6% of adjusted capital. Borrowing at 4.75% to buy stock at book is a different bet from buying it with underwriting earnings.

  5. Favourable prior-year development stays above 10% of pre-tax income for a third year, or the 2021-22 excess-casualty years are strengthened.

    Details

    FY2025 took $427m of releases, 11.0% of pre-tax income and 3.6x the FY2024 figure, in the year the 10%+ core operating ROE target came due; H1 2026 leans on "more favorable prior year reserve development" again. An analyst asked specifically about accident years '21 and '22 on 2026-08-07 and was answered about '16 and '23. The Q3 2026 10-Q's development table settles it.

  6. The FY2026 low-to-mid-teens premium-growth guide is missed without having been withdrawn, or a second consecutive call passes without management reaffirming it when asked.

    Details

    It was given on 2026-05-01 [4] and left unaddressed on 2026-08-07 [42]; H1 came in at +13%.

  7. Capital return exceeds operating cash flow in FY2027, the first year without Corebridge proceeds.

    Details

    Five years of buybacks and dividends ran at 122% of operating cash flow because assets were being sold; that source is nearly exhausted. If the pace is held anyway it is being borrowed.

  8. A 2027 Investor Day number is restated, re-based or dropped.

    Details

    — the 20%+ EPS CAGR, the 10-13% ROE band, the sub-30% expense ratio or the 94% Global Personal combined ratio. Three targets have already gone quiet or unwound on this management (the President appointment, the 15-20% leverage target, the growth guide); a fourth would make it the pattern rather than the exception.

How far along

7 of 10 steps done · 3 done in a group review

Next step

watch Item 5.02 8-Ks between now and the Q3 call. Two of the company's most senior people left in fifteen days — the Executive Chair who built the restructuring and the CEO of General Insurance, which after the Corebridge exit is effectively the whole company. The filings say no disagreement and the stock did not move (−0.9% across both announcements). A third departure at that level, or a named external hire into General Insurance, is the fact that decides whether this is an orderly succession or the start of something. The Q3 call on ~2026-11-04 is the first time anyone gets to ask.

Needs your decision

S6 product test / expert call, S7 conviction.

Failed checks

none. S0/S4/S6/S7 are not closed/closed by carry-forward. S3 is closed with the one-transcript deviation named above — narrowed on 2026-09-24, which read all eight archived calls for S8. S8 is closed (partial): new team, thin record. S2's capital-allocation verdict is neutral and S3's earnings-quality check carries 3 flags.

Who runs it

The Front Office

Step by step

  1. S0How it does as AI gets cheapDone in a group review

    carry-forward (cluster: Insurance — Stage 0 re-score, 2026-09-22 §5)

    lens 3/4 under the SKILL.md §4.3 risk-carrying variant, which supersedes the.

    ⚠️ 2/4 the AI-surplus four gave it in Insurance & reinsurance — cluster pass, 2026-09-22. Reserve honesty 1 (favourable in 4 of 5: +78, then −356 / −234 / −120 / −427 $m; paid-to-incurred 0.931-0.990, no drift; caveat that it is net of the adverse development cover); edge survives the cycle 1 (accident year combined ratio as adjusted 91.0 / 88.7 / 87.7 / 88.2 / 88.3 — below 95 in 5 of 5, and AIG publishes the measure itself); capital theirs 1 (1-in-250 all-peril PML $2,500m net, $1,975m after tax = 4.8% of equity — the strongest catastrophe disclosure of the seven); growth that does not dilute 0 (buybacks $6,652m and $5,836m in 2024-25 against $3,273m and $3,314m of operating cash flow, while premiums and reserves shrank). Liquidity: ADV $314.3m/day (21 sessions to 2026-09-22, market data NYSE:AIG). ⚠️ THIS PASS OVERTURNS THE INPUTS TO THE 0 — see S2. Glyph records the lens ran, not its verdict.

  2. S1Where the idea came fromDone

    watchlist (owner, Apple Stocks, 2026-09-20). No Tier 0 promotion; second member of this slot.

  3. S2Quick deal-breaker checkDone

    FY2025 revenue $26,775m, net income $3,096m, net earned premium $23,751m, equity $41,139m, reserves $70,666m, operating cash flow $3,314m (XBRL, 10-K 0000005272-26-000023)

    Full notes

    H1 2026: diluted EPS $3.18 (H1 2025 $3.13) on 537.8m diluted shares (588.5m), BVPS $77.39, adjusted tangible BVPS $72.18, total debt $9bn at 17.6% of adjusted capital. Five-year diluted share count 657.3m (2024) → 570.3m (2025) → 524.7m outstanding at 2026-06-30: about −34% from 2024. Survives easily; there is no solvency question here. Two inputs to the Stage 0 zero have reversed in 2026 and the scorecard says so rather than carrying the old reading: (1) General Insurance net premiums written +9% in Q2 and +15% over H1 2026 ($13,115m vs $11,406m) — the shrink is over. (2) the buyback is now inside operating cash flow — $1,153m of purchases in H1 2026 against $4,007m a year earlier, with $1,871m of operating cash flow and $504m of dividends. A re-score belongs in a cluster file, not here; this is the evidence for one. · capital allocation (2026-09-24, dossier-2026-09-24.md): neutral. FY2021-25 buybacks $23,241m + dividends $5,040m = $28,281m against $23,187m of operating cash flow — 122% of it. The gap was funded by taking the company apart: $7,838m of tagged divestiture proceeds, the Corebridge sell-downs (to 10.1% at 2025-12-31, full exit guided for 2026) and the deconsolidation itself (assets $539bn -> $161bn), with long-term debt cut 69% to $9,191m. The buybacks were done well: $20,649m across 2022-25 against a 280.5m fall in shares outstanding = ~$74/share against BVPS of $76.44 — a third of the company retired at or below book, which is most of why core operating ROE went 9.1% -> 11.1%. But five years of book value went nowhere: BVPS $77.02 (2020-12-31) -> $76.44 (2025-12-31), -0.8%, and +8.7% (1.7%/yr) adding the $7.27 of dividends declared — AOCI and the deconsolidation distort it, and it is still the number. No goodwill impairment in the window; goodwill fell only through disposals.

    ⚠️ The liquidation is nearly finished: from 2027 the capital return has to come out of a $3.3bn operating cash flow, not out of asset sales.

  4. S3Reading the company's reportsDone

    (gate-minimum)

    FY2025 10-K Item 1A Risk Factor Summary (three disclosed risks named below, in management's own ordering) and PwC's report.

    Full notes

    Q2 2026 10-Q MD&A (General Insurance and the three segment tables, Note 13, the cash flow statement); the 2026-03-31 DEF 14A ownership table. the 2026-09-02 and 2026-09-16 Item 5.02 8-Ks; the 2026-09-16 424B2 for the EUR 1.125bn note issue; the Q2 FY2026 call transcript; and a full efts.sec.gov red-flag sweep — zero "changes in accountants", one "non-reliance" hit that is a 10-Q exhibit, "restatement" and "material weakness" hits all clawback exhibits, award agreements or ICFR boilerplate; no NT 10-K, no 10-K/A or 10-Q/A in the index. PwC, PCAOB 238, sole CAM the loss reserves ($41.8bn net at 2025-12-31). Deviation: one transcript, not two. EQ: 3 flags (2026-09-24, dossier-2026-09-24.md) — (1) restructuring charges excluded from the headline in six consecutive years ($435m, $433m, $423m, $356m, $745m, $439m for 2020-25. $2,396m across the five-year table), one of twenty categories the release names as excluded from adjusted pre-tax income. (2) cash conversion below 1.0x and falling — operating cash flow $6,243m (2023) -> $3,273m (2024) -> $3,314m (2025) while AATI went $3,254m -> $4,044m and AATI/share rose 43% to $7.09, so OCF/AATI is 1.92x -> 1.01x -> 0.82x (part structural, Corebridge's cash left with the deconsolidation). (3) favourable prior-year development tripled in the year the ROE target came due — $120m (3.1% of pre-tax income) in FY2024 to $427m (11.0%) in FY2025, on pre-tax income otherwise flat at $3,870m -> $3,879m. Upper-bound arithmetic: without the extra $307m pre-tax, core operating ROE would be ~10.4% rather than 11.1% — still above target, with 40bp of headroom instead of 110bp. It continues into H1 2026. Mild: DAC/NPE 8.3% -> 8.8% -> 8.9% on flat earned premium. Clean: SBC 0.77% of revenue and falling. FY2025 accruals -0.14% of average assets.

  5. S4Rivals and suppliersDone in a group review

    carry-forward (cluster: Insurance & reinsurance — cluster pass, 2026-09-22 §2)

    large-account commercial, absorbs loss, paid on risk margin, "worst-exposed segment" to soft pricing.

  6. S5Who owns the stockDone

    13F by manager at 2026-06-30 (BlackRock 48.9m sh / $3,647m. Wellington 30.3m, +1.96m, the largest active add; State Street 25.1m. Harris Associates 20.4m, −0.24m. Hotchkis & Wiley +1.61m. Diamond Hill −2.34m.

    Full notes

    Ameriprise −6.73m), 13D/G census (no activist, no strategic block). Insider ownership ≈0.6% — 22 directors and executive officers hold 3,238,630 shares and no individual holds 1%, the thinnest alignment of any name in this cluster. Short interest 8,709,587 shares at the 2026-08-31 settlement (~1.7% of shares out, 3.2 days to cover), rising three settlements running from 7.42m. Buyback $2.6bn remaining at 2026-07-31; ~2m shares / ~$195m in July alone. Shelf: AIG files off an existing shelf (File No. 333-277075) and drew EUR 1.125bn on it on 2026-09-16.

    ⚠️ the Q2-26 13F quarter is incompletely ingested (764 filers vs 1,108) — both Vanguard entities, Geode, Franklin, GQG, Morgan Stanley, both Capital entities, Dimensional and Northern Trust have no filing in it, so no "exit" may be read from an absence.

    ⚠️

    ⚠️ the Sixth Street row is NOT a holding — sixth issuer, see the log.

  7. S6What workers and customers sayDone

    carry-forward (cluster: Insurance & reinsurance — cluster pass, 2026-09-22 §4)

    two claim rows from the 2026-09-07 Real Eisman Playbook: "I am recommending AIG" (Tunis) and "And full disclosure, everybody, I own AIG" (Eisman). ⚠️ BOTH PRE-DATE THE 2026-09-01 AND 2026-09-16 8-Ks and neither speaker had seen them.

    Full notes

    Product test / expert call open OPEN (user)

  8. S7The case for it, and pricesDone in a group review

    carry-forward (cluster: Insurance & reinsurance — cluster pass, 2026-09-22 §5)

    "self-help is real and nearly complete; the residual is a price-taker in the softest segment", kill criterion "buyback pace falls below the pace of premium decline". That criterion has now half-fired in a way it did not anticipate — the buyback fell 71% while premiums turned UP. Replaced below.

    Full notes

    Conviction open OPEN (user) · valuation (bear/base/bull; moved from S8 2026-09-23): see the valuation table below. P/B 0.97 and P/adjusted-tangible-book 1.04 at $75.18 (2026-09-22) — the only P&C name in this cluster at or below book. Base ~$92, bear ~$60, bull ~$130.

  9. S8Do the bosses keep their word?Done

    (partial)

    new team, thin record (2026-09-24, dossier-2026-09-24.md). 9 rows: 4 beat, 2 met, 3 withdrawn — and seven of them belong to Peter Zaffino, who left the Board on 2026-09-15. Eric Andersen, CEO since 2026-06-01 (8-K 0000005272-26-000042), has completed no scored row; everything he has committed to comes due in 2027.

    Full notes

    Zaffino's record was conservative: AIG Next beaten and early ($530m actioned against "at least $500m", delivered by Q2 2025). FY2025 core operating ROE 11.1% against a 10%+ target; buybacks $5.8bn at the top of a $5-6bn range; the 2025 dividend +12.2% against a "in line with the share-count decrease" promise; the 2026 dividend +11% against +10% committed; share count 570.3m inside a 550-600m target.

    ⚠️ None of those targets required AIG to grow — they are expense, capital-return and return-on-a-shrinking-equity-base targets, which is what a company dismantling a conglomerate can promise.

    ⚠️ Three withdrawn rows share a shape: John Neal was named President effective 2025-12-01 by 8-K (0000005272-25-000068) and "will no longer be joining" four months later (0000005272-25-000166); the 15-20% leverage target has not been "reiterated ... in quite some time" (CFO, call 2026-08-07 [44]); and the FY2026 low-to-mid-teens premium-growth guide was asked about directly on 2026-08-07 [41] and neither reaffirmed nor revised [42]. Consequence for S7: the Investor Day numbers (20%+ EPS CAGR, 10-13% ROE, sub-30% expense ratio, 94% Global Personal, all by 2027) are not the old team's record and must not inherit its credibility — take the base case from the H1 2026 run rate (core operating ROE 11.6%, trailing expense ratio 30.7%), not from the 2027 endpoints.

  10. S9What to watch nextDone

    trigger written; next earnings ~2026-11-03/04 (DERIVED from filing cadence: Q3 results 8-Ks landed 2024-11-04 and 2025-11-04. Equibles' IR calendar was not queried for this name and market data's needs a paid plan)

    Full notes

    Tier 0 EDGAR sweep covers the feed (cik 5272), and on this name Item 5.02 is the form to watch, not the 10-Q.

Sources and notes

Sources 10 documents
Short nameDocumentWhere it came fromRead on
10K-2025AIG 10-K for FY2025, filed 2026-02-12https://www.sec.gov/Archives/edgar/data/5272/000000527226000023/aig-20251231.htm2026-09-23
10Q-Q2-26AIG 10-Q for the quarter ended 2026-06-30, filed 2026-08-07https://www.sec.gov/Archives/edgar/data/5272/000000527226000076/aig-20260630.htm2026-09-23
8K-ZAFFINOAIG 8-K filed 2026-09-02 (event 2026-09-01), Item 5.02 — Executive Chair steps downhttps://www.sec.gov/Archives/edgar/data/5272/000000527226000088/aig-20260901.htm2026-09-23
8K-HANCOCKAIG 8-K filed 2026-09-16, Item 5.02 — CEO of General Insurance retireshttps://www.sec.gov/Archives/edgar/data/5272/000000527226000091/aig-20260916.htm2026-09-23
424B2-EURAIG 424B2 filed 2026-09-16 — EUR 1.125bn of 2031 and 2036 notes, settling 2026-09-24https://www.sec.gov/Archives/edgar/data/5272/000110465926108322/tm2623943-3_424b2.htm2026-09-23
DEF14A-26AIG DEF 14A filed 2026-03-31 (director and officer ownership table)https://www.sec.gov/Archives/edgar/data/5272/000000527226000039/aig-20260331.htm2026-09-23
13F-Q2-26Institutional holders by manager, 13F report date 2026-06-30 (ingestion incomplete — see the edge note)python ledger/research/tools/threespread.py holders AIG2026-09-23
CALL-Q2-26AIG Q2 FY2026 earnings call, 2026-08-07, speaker-labelled transcriptEquibles GetEarningsCallTranscript(ticker=AIG, fiscalYear=2026, fiscalQuarter=2) — https://equibles.com/stocks/aig/earnings-calls2026-09-23
CLUSTER-INS-RESCOREInsurance Stage 0 re-score, 2026-09-22, slot cluster-insurance-rescore (§5 is this name)ledger/research/_clusters/cluster-insurance-rescore-2026-09-22.md2026-09-23
CLUSTER-INSInsurance & reinsurance cluster pass, 2026-09-22, slot cluster-insurance (S4 / S6 / S7)ledger/research/_clusters/cluster-insurance-2026-09-22.md2026-09-23
Links to other companies 19 links
context 6holds 3key person risk 2officer of 2acquired 2director of 1auditor of 1issuer of 1governance structure 1
FromLinkToDateWhere it says so
Peter Zaffino (Executive Chair until 2026-09-15; Senior Advisor to the CEO thereafter)key person riskAIG2026-09-018K-ZAFFINO — Item 5.02, first and second paragraphs
John Rice (Lead Independent Director; Chair of the Board from 2026-09-15)director ofAIG2026-09-158K-ZAFFINO — Item 5.02, first paragraph
Jon Hancock (EVP and CEO, General Insurance, until 2026-12-31)key person riskAIG2026-09-168K-HANCOCK — Item 5.02, sole paragraph
Eric Andersen (President and CEO)officer ofAIG2026-09-168K-HANCOCK — Item 5.02, sole paragraph — 'Mr. Hancock will report to Eric Andersen, President & Chief Executive Officer of AIG'
Keith Walsh (CFO)officer ofAIG2026-08-07CALL-Q2-26 — Q2 FY2026 call, prepared remarks, second speaker
PricewaterhouseCoopers LLP (PCAOB ID 238)auditor ofAIG2026-02-1210K-2025 — Report of Independent Registered Public Accounting Firm, Critical Audit Matters — 'Valuation of Insurance Liabilities - Unpaid Losses and Loss Adjustment Expenses (Loss Reserves), Net of Reinsurance'
AIGcontextAIG2026-06-3010Q-Q2-26 — MD&A, General Insurance underwriting results table
AIGcontextAIG2026-07-3110Q-Q2-26 — Condensed consolidated statement of cash flows, financing activities; Note 13, Repurchase of AIG Common Stock
AIGissuer ofEuronext Dublin (Global Exchange Market)2026-09-16424B2-EUR — Prospectus supplement cover and 'Use of Proceeds' summary, p. S-3
AIGacquiredCorebridge Financial, Inc.2026-05-31CALL-Q2-26 — Q2 FY2026 call, CFO prepared remarks, capital management paragraph
AIGacquiredEverest Group (insurance operations in Colombia; US renewal rights)2026-08-07CALL-Q2-26 — Q2 FY2026 call, CEO prepared remarks, geographic presence paragraph
Eric Andersen (President and CEO)contextAIG2026-08-07CALL-Q2-26 — Q2 FY2026 call, answer to Mayor Shields (KBW) on social inflation
AIGcontextAIG2026-06-30CALL-Q2-26 — Q2 FY2026 call, CFO prepared remarks, prior-year development paragraph
BlackRock, Inc.holdsAIG2026-06-3013F-Q2-26 — 13F-HR, accession 0002012383-26-003238 (holders table, row 1)
Wellington Management Group LLPholdsAIG2026-06-3013F-Q2-26 — 13F-HR, accession 0000902219-26-000311 (holders table, row 2)
Harris Associates L.P.holdsAIG2026-06-3013F-Q2-26 — 13F-HR, accession 0000813917-26-000042 (holders table, row 4)
Sixth Street Partners Management Company, L.P.contextCharles Schwab Investment Management, Inc.2026-06-3013F-Q2-26 — threespread.py holders AIG — identical-share-count warning, rows 14 and 15
AIGgovernance structureAIG2026-03-31DEF14A-26 — Ownership of common stock table and footnote
AIGcontextAIG2026-02-1210K-2025 — Item 1A, Risk Factor Summary, 'Employees and Competition', p.12
History 3 entries
  • 2026-09-23 — first pass, slot names-insurance-2, model-drafted unattended by the cloud research routine. S1, S2, S3 (gate-minimum), S5 and S9 closed; S0 carried forward from Insurance — Stage 0 re-score, 2026-09-22 §5 and S4/S6/S7 from Insurance & reinsurance — cluster pass, 2026-09-22; S8 modelled and left open. Run finding: two Item 5.02 8-Ks fifteen days apart removed the Executive Chair from the Board and put the CEO of General Insurance on a retirement path, against a stock that moved −0.9% across both and scuttlebutt written between them that reflects neither. Second finding: both inputs to the Stage 0 test-4 zero reversed in H1 2026 — buybacks $1,153m (from $4,007m) inside $1,871m of operating cash flow, and General Insurance net premiums written +15%. A cluster re-score should carry that, not this file. Third: management says social inflation "hasn't" moderated and is not pricing for it, in the same week Allstate's CEO attributed large releases to tort reform — two carriers, opposite readings, one booking releases on it. Fourth: EUR 1.125bn of new notes on 2026-09-16 refinance EUR 821m of 1.875% paper due June 2027 at 4.250%/4.750%. Tooling: the Sixth Street / Charles Schwab IM 13F defect appears for the sixth issuer (4,343,856 shares); row discarded. Also recorded: the Equibles transcripts for both names in this slot mis-assign speaker labels (analyst names on management turns and vice versa) — two for two, so every attribution from that source needs an internal check before it is carried. Evidence and access limits in dossier-2026-09-23.md. Nothing here is human-verified.
  • 2026-09-23 — stage redefinition, owner decision (WATCHLIST_PLAN.md §25): S8 is now Management credibility and is not yet run for this name; the old S8 valuation figures moved to the valuation: line under S7, and any proposed buy price or position size was removed (the pipeline no longer drafts either). No research was redone. S2 capital allocation and S3 earnings quality are also pending, in the same backfill slot.
  • 2026-09-24 — backfill, slot backfill-2026-09-23-5, model-drafted unattended by the cloud research routine. S8 management credibility (9 rows, 4 beat / 2 met / 3 withdrawn — new team, thin record, ✅ partial), S2 capital allocation (neutral) and S3 earnings quality (3 flags) are now written. Evidence: EDGAR only (the 2026-02-10 Q4/FY2025 release, ten Item 5.02 8-Ks, XBRL companyfacts) plus all eight archived market data transcripts. 0 Equibles calls, 0 market data calls, 0 3spread calls. No other stage was redone. Run finding 1 — the departures have a price. Seven of the nine scored promises were made by Peter Zaffino, who ceased to be an officer or director on 2026-09-15; Eric Andersen, CEO since 2026-06-01, has completed none. The 2027 Investor Day targets should not inherit the old team's credibility. Run finding 2 — what the old team actually promised. Every 2025 target was an expense, capital-return or return-on-a-shrinking-base target. None of them required AIG to grow. They were all hit, several early and above the number; that is a real record and it is a record of restructuring. Run finding 3 — the reserve release and the ROE target arrived together. Favourable prior-year development went $120m (3.1% of pre-tax income) to $427m (11.0%) in FY2025, the year the 10%+ core operating ROE target came due, on pre-tax income otherwise flat. Upper-bound arithmetic puts the target at ~10.4% without the step-up — still met, with 40bp of headroom rather than 110bp. Run finding 4 — a President who never arrived. John Neal was named President by 8-K effective 2025-12-01 and "will no longer be joining" four months later. Scored as withdrawn: a dated operational commitment made in a filing and not kept. Tooling: market data's archived transcript for the call of 2024-11-05 labels the CFO turn "Sabra Purtill", two weeks after EDGAR says Keith Walsh became CFO (8-K 0000005272-24-000101, effective 2024-10-21). Labels are reported as read; roles were taken from EDGAR, as SKILL.md §S8 requires. Nothing here is human-verified.
Full notes, 24 Sep 2026 18 min read

AIG — dossier, 2026-09-24 (backfill: S8, S2 capital allocation, S3 earnings quality)

Slot backfill-2026-09-23-5. Model-drafted unattended by the cloud research routine; nothing here is human-verified. This file holds only the three checks added on 2026-09-23 (WATCHLIST_PLAN.md §25). Every other stage stands as written in dossier-2026-09-23.md. Anything from background knowledge is marked [background].

Sources this pass. EDGAR, User-Agent ddboard-research gorkem@celebico.com (CIK 5272): the Q4/FY2025 earnings release 8-K Ex. 99.1 of 2026-02-10 (q42025earningsrelease.htm); the Item 5.02 8-Ks of 2024-09-18, 2025-07-21, 2025-08-18, 2025-10-03, 2025-11-14, 2026-01-06, 2026-03-16, 2026-04-27, 2026-09-02 and 2026-09-16; XBRL companyfacts for CIK 5272. Transcripts entirely from the archive ledger/research/_transcripts/AIG/ (market data, eight quarters FY2024-Q3 through FY2026-Q2, call dates from _transcripts/index.json). Equibles: 0 calls. market data: 0 calls. 3spread: 0 calls. Nothing was fetched from the open web.

Quote discipline: transcript quotes were located by exact substring in the archived JSON bytes and copied from them; filing quotes were matched against a whitespace-collapsed text extraction of the EDGAR HTML. Turn numbers [n] are newline positions in the market data content string.

S8 — management credibility

The headline is not the record. It is that the person who made the record has left.

Who promised, and who is still here

PersonRole, and its dates on EDGARStatus
Peter ZaffinoChairman & CEO through 2026-05-31; Executive Chair 2026-06-01 → 2026-09-15; then Senior Advisor to the CEO, off the Boarddeparted the Board 2026-09-15 — 8-K 0000005272-26-000088 Item 5.02: "On September 1, 2026, Peter Zaffino notified the American International Group, Inc. (the "Company") Board of Directors (the "Board") that he will step down as Executive Chair and a member of the Board, effective September 15, 2026". Made every target in the table below
Eric AndersenPresident & CEO Elect from 2026-02-16; President & CEO and director from 2026-06-01 (8-K 0000005272-26-000042 Item 5.02). Previously President of Aon plc 2020–2025in post four months. Has completed no scored row
Keith WalshEVP & CFO from 2024-10-21 (8-K 0000005272-24-000101 Item 5.02)in post throughout the promise window
Sabra PurtillEVP & CFO until Walsh's start, then advisor (same 8-K)departed the role 2024-10. Discrepancy worth recording: market data labels the CFO turn on the call of 2024-11-05 as "Sabra Purtill" [3], two weeks after EDGAR says Walsh became CFO. The label is reported as read; the role is taken from EDGAR, as SKILL.md §S8 requires
Jon HancockEVP & CEO, General Insurance; spoke on both 2026 callsretiring 2026-12-31 — 8-K 0000005272-26-000091 Item 5.02, announced 2026-09-16
John Nealnamed President effective 2025-12-01 (8-K 0000005272-25-000068, announced 2025-07-16)never joined — see row 8
Claude WadeEVP, Chief Digital Officer & Global Head of Business Operations & Claimsstepped down 2025-12-31 for health reasons (8-K 0000005272-25-000098)

So: the CEO who set every target left the board nine days before this pass ran, the head of General Insurance leaves at year end, a President appointment was announced and then unwound, and the Chief Digital Officer left. The CFO is the continuity.

Say vs do (guides made 2024-11-05 → 2026-08-07)

#Promise (period · metric · figure)Made where · by whomResult · locatorScore
1AIG Next: at least $500m of run-rate savings, fully realised in 2025call 2024-11-05 [2] · Peter Zaffino (Chairman & CEO): "In 2025, we expect to fully realize the $500 million in savings from AIG Next." Path recorded: "We exited 2024 achieving $450 million in run rate savings as part of the program and we expect the remaining benefits to be realized in the first half of 2025" (call 2025-02-12 [2])call 2025-08-07 [2] · Zaffino: "We actioned over $530 million of annual run rate expense savings with over $500 million realized through the second quarter." Delivered above the number and a half-year earlybeat (departed)
2Parent expense ≈ $350m a year going forwardcall 2024-11-05 [2] · Zaffino: "we've established a new definition of parent expense to exclusively reflect costs related to being a global regulated public company and expect those costs to be around $350 million going forward"$360m — release 2026-02-10, Other Operations table, "Corporate and other general operating expenses ... (623) (360)" for FY2024 and FY2025. A 42% cut, landing $10m (2.9%) above "around $350 million"met (departed)
3FY2025 · core operating ROE of 10%+call 2024-11-05 [2] · Zaffino: "we expect to deliver a 10% core operating ROE for the full-year 2025", restated on every subsequent call11.1% — release 2026-02-10, quoting Zaffino: "Core Operating ROE of 11.1% was above our 10% plus target for 2025." Confirmed in the Selected Financial Data table, "Core operating ROE (b÷f) ... 11.1 %"beat (departed) — but see the reserve note in S3
4Share count into a target range of 550–600mcall 2024-11-05 [2] · Zaffino: "The current authorization will bring us within our target share count range of 550 million to 600 million shares."570.3m weighted-average diluted for FY2025 — release 2026-02-10, "Weighted average common shares outstanding - diluted ... 657.3 570.3". Inside the range, and the target was then re-set to 500–550m (call 2025-05-02 [2])met (departed)
5FY2025 · $5–6bn of share repurchasescall 2025-05-02 [2] · Zaffino: "We expect to repurchase a total of $5 billion to $6 billion of shares in 2025"; Walsh guided the high end on 2025-08-07 [3]$5.8bn — release 2026-02-10: "Returned $6.8 billion of capital to shareholders, including $5.8 billion of share repurchases and approximately $1.0 billion of dividends"met (high end)
62025 dividend up in line with the share-count decreasecall 2025-02-12 [2] · Zaffino: "We are committed to reviewing our dividend annually and anticipate that we will increase our dividend in 2025 in line with the decrease in our share count over the past year, subject to AIG board approval."+12.2% — XBRL CommonStockDividendsPerShareDeclared $1.56 (CY2024) → $1.75 (CY2025) — against a 9.4% fall in diluted shares over 2024 (725.2m → 657.3m). Delivered more than "in line"beat (departed)
72026 dividend +10% (Investor Day 2025 commitment)reaffirmed on the call of 2026-05-01 [3] · Eric Andersen (President & CEO Elect), listing the Investor Day guidance: "supporting the increase in our dividend by 10% in 2026"+11%, announced the same day — call 2026-05-01 [2] · Zaffino: "the AIG Board of Directors approved an 11% increase in our quarterly dividend to $0.50 per share starting in the second quarter of 2026"beat
8John Neal to serve as President, effective 2025-12-01 — a dated operational commitment, announced by 8-K8-K 0000005272-25-000068 Item 5.02, filed 2025-07-21: "On July 16, 2025, American International Group, Inc. (the "Company") announced that John Neal, age 60, will serve as President, effective December 1, 2025."did not happen — 8-K 0000005272-25-000166 Item 5.02, filed 2025-11-14: "AIG... has reached a mutual agreement with John Neal, who had been named to the position of President effective December 1, 2025, that he will no longer be joining the Company due to personal circumstances." Two months later AIG hired a President and CEO from outside insteadwithdrawn (departed)
9Financial leverage target of 15–20%the standing target [background: set in the 2022–2023 deleveraging]; it is the target an analyst had to ask about because it stopped being repeatedquietly retired — call 2026-08-07 [44] · Keith Walsh (EVP & CFO), to Elyse Greenspan: "We haven't given -- we haven't reiterated that 15% to 20% in quite some time. Obviously, we're at 17.6%. We're within that range." The company is inside the range and has stopped naming itwithdrawn

Not scored (open). The 2025 Investor Day package, reaffirmed verbatim by the incoming CEO on the call of 2026-05-01 [3] — "delivering operating EPS compound annual growth of over 20% over the 3 years ending 2027; driving core operating ROE of 10% to 13% through 2027; improving General Insurance's expense ratio to less than 30% by 2027; ... and achieving improvement in Global Personal Insurance combined ratio to 94% by 2027." Progress where it is measurable: core operating ROE 11.6% for H1 2026 (call 2026-08-07 [2]); trailing-twelve-month expense ratio 30.7% at 2026-06-30 (call 2026-08-07 [3]); Global Personal running at or better than the 94% target (call 2026-08-07 [34], Pablo Singzon, unchallenged). Also open: FY2026 net premiums written growth "low to mid-teens" (call 2026-05-01 [4], Walsh) with H1 at +13%; a full exit from Corebridge in 2026 (ownership already down to 10.1% at 2025-12-31, release 2026-02-10); and "approximately $3 billion of subsidiary dividends in 2025" (call 2025-08-07 [3], Walsh) — not checked this pass, it needs the 10-K's parent-company liquidity discussion.

Tally: 9 rows — 4 beat, 2 met, 3 withdrawn. By team: Peter Zaffino, who has left, owns rows 1–6 and 8 — 4 beat, 2 met, 1 withdrawn. Row 7 he announced and Andersen had already reaffirmed; row 9 is Walsh's. Eric Andersen, CEO since 2026-06-01, owns no completed row at all. Every Investor Day number he has committed to comes due in 2027, after this scorecard's horizon.

Dodge-log roll-up (§3.1). No dodge log existed for this name; it is opened below from the two most recent calls.

  1. The company's own FY2026 growth guide, asked about and not addressed. Call 2026-08-07 [41] · Elyse Greenspan: "I don't think you guys updated -- gave an update on the premium growth guide, right, which was low to mid-teens for the year... do you think you're at the low end or maybe a bit below that as we think about potential growth in the second half?" Eric Andersen answered with posture — "we're not going to chase growth blindly that we want to make sure we maintain our underwriting standards" [42] — and neither reaffirmed nor revised the number. The guide was given on the previous call in terms: "For full year 2026, we continue to expect low to mid-teens net premium written growth in General Insurance" (call 2026-05-01 [4], Walsh). A dodge on management's own guidance is the strongest signal this stage produces (SKILL.md §3.1), and it is the new CEO's first one.
  2. Sizing the Convex quota share. Call 2026-08-07 [28] · Rowland Mayor asked him to "size the premium contribution from the Convex quota share in the quarter... since there's built-in growth from that over the next few years"; the answer gave organic growth "somewhere I would call low to mid-single digits" and then "with the Everest transaction, it gets a little difficult to track as we start to go client by client" [29] — no figure for the transaction asked about. With growth increasingly coming from renewal-rights and quota-share deals, the split between bought and organic premium is the number that decides whether the 13% is repeatable.
  3. Reserve adequacy by accident year. Call 2026-08-07 [11] · Meyer Shields asked specifically about accident years '22 and '21 — "Is there significant IBNR there that would also need to be reviewed?" After a preamble from the CEO on process [12], Walsh answered about '16 and '23 and about ranges of estimates [13]. The years asked about were not addressed. Logged as a note this quarter; if it recurs it becomes a finding, because it is the same book that S3's prior-year-development flag is about.

Verdict — new team, thin record; the old team's record was conservative. Under SKILL.md §S8 this closes ✅ (partial): most of the scored record belongs to Peter Zaffino, who is no longer an officer or a director, and the CEO in post has completed nothing. What Zaffino left behind is a good record — the $500m cost programme beaten and delivered early, a 10%+ ROE target hit at 11.1%, the buyback and dividend commitments met at or above their ranges — and it is worth saying that plainly. Two things temper it and both are structural rather than a matter of effort: none of the 2025 targets required AIG to grow; they were expense, capital-return and return-on-a-shrinking-equity-base targets, which is what a company dismantling a conglomerate can promise. And the three withdrawn rows share a shape — a President appointment that unwound, a leverage target that stopped being repeated, and now a growth guide that went unaddressed when asked. Consequence for S7: the Investor Day numbers (20%+ EPS CAGR, 10–13% ROE, sub-30% expense ratio by 2027) are not the old team's record and should not inherit its credibility. Model them as an unproven management's plan and take the base case from the H1 2026 run-rate — 11.6% core operating ROE, 30.7% trailing expense ratio — rather than from the 2027 endpoints.

S2 — capital allocation

All figures $m, calendar fiscal years, from XBRL companyfacts (CIK 5272) unless stated: OCF NetCashProvidedByUsedInOperatingActivities, buybacks PaymentsForRepurchaseOfCommonStock, dividends PaymentsOfDividendsCommonStock, SBC AllocatedShareBasedCompensationExpense, debt ProceedsFromIssuanceOfLongTermDebt / RepaymentsOfLongTermDebt, divestitures ProceedsFromDivestitureOfBusinessesNetOfCashDivested, restructuring RestructuringCharges.

FY2021FY2022FY2023FY2024FY20255y total
Operating cash flow6,2234,1346,2433,2733,31423,187
Buybacks2,5925,2002,9616,6525,83623,241
Common dividends1,0839829971,0029765,040
Divestiture proceeds4,68302,56858707,838
Debt raised1077,4777426611,24110,228
Debt repaid4,1479,4552,304n/tn/t≥15,906
Stock-based compensation2782881992112071,183
Restructuring charges4334233567454392,396
Long-term debt (year end)30,16327,17910,6068,9229,191−69%
Diluted weighted-average shares (m)864.9787.9725.2657.3570.3−34.1%
Shares outstanding (m)818.7734.1688.8606.1538.2−34.3%

Where the cash went — and where it came from. Buybacks plus dividends over five years are $28,281m against $23,187m of operating cash flow: 122%. The gap was not funded by earnings. It was funded by taking the company apart: $7.8bn of tagged divestiture proceeds, the Corebridge IPO and successive sell-downs (ownership down to 10.1% at 2025-12-31 after a Q4 sale for "approximately $1 billion", release 2026-02-10, with a full exit guided for 2026), and the deconsolidation itself — total assets fell from $539bn at 2023-12-31 to $161bn at 2024-12-31 (Assets, XBRL). Long-term debt was cut 69% in the same window. This is a five-year liquidation of a conglomerate, executed well, and it is nearly finished. The question S7 has to answer is what the capital-return run rate looks like in 2027, when Corebridge is gone and buybacks have to come out of a $3.3bn operating cash flow rather than out of asset sales.

Buybacks at what price — and this is the good news. 2022–2025 buybacks of $20,649m against a 280.5m fall in shares outstanding (818.7m at 2021-12-31 → 538.2m at 2025-12-31) implies ≈$74 a share before allowing for shares issued into compensation, which makes the true average a little lower still. Book value per share was $76.44 at 2025-12-31 and $70.16 a year earlier (release 2026-02-10). AIG has been retiring a third of its shares at or below book value. That is the single most value-accretive thing in this file, and it is a large part of why core operating ROE rose from 9.1% to 11.1% — the denominator shrank faster than the numerator grew.

What the acquisitions were worth. No goodwill impairment is tagged in the window (GoodwillImpairmentLoss is empty for 2021–2025; GoodwillImpairedAccumulatedImpairmentLoss is flat at $3,410m from 2022, i.e. no new impairment). Goodwill fell $4,056m → $3,435m through disposals, not write-downs: GoodwillWrittenOffRelatedToSaleOfBusinessUnit $369m (2023) and $22m (2024). The recent deployments are the other way round — renewal-rights and quota-share transactions (Everest, Convex) and a $2bn commitment over three years into Onex's asset-management platform (call 2025-11-05 [2], Zaffino). Management says these "should enable us to grow, deliver earnings and improve ROE without adding complexity" (release 2026-02-10, Zaffino) and expects them to be "earnings, EPS and ROE accretive 1 year after closing" (call 2025-11-05 [2]) — a dated commitment that comes due in late 2026 and is not yet scorable.

Returns (insurer form — ROE and book value per share compounded with dividends, per SKILL.md).

FY2021FY2022FY2023FY2024FY2025
Net income (loss) to common ($m)10,33810,1983,614(1,426)3,096
Average AIG equity ($m)66,15953,46343,16143,93641,830
GAAP ROE15.6%19.1%8.4%(3.2)%7.4%
Core operating ROE (company measure)———9.1%11.1%

The GAAP line is noise: 2021 and 2022 carry Corebridge IPO and divestiture gains, 2024 carries the deconsolidation loss. The company's own core operating ROE is the usable series and it goes the right way — 9.1% → 11.1%, and 11.6% for H1 2026 (call 2026-08-07 [2], Andersen).

Book value per share, which is where the five years actually land. AIG's reported BVPS of $76.44 at 2025-12-31 (release 2026-02-10) equals StockholdersEquity $41,139m over 538.2m shares outstanding, so the same arithmetic gives $77.02 at 2020-12-31 ($66,362m over 861.6m). That is −0.8% over five years. Adding the $7.27 of dividends declared across 2021–2025 (XBRL CommonStockDividendsPerShareDeclared / ...CashPaid: 1.28, 1.28, 1.40, 1.56, 1.75) takes the five-year total return on book to +8.7%, i.e. 1.7% a year. Two honest qualifications: AOCI took a large bite in 2022 as rates rose and has been giving some back (book value "up 4% from the prior year quarter, reflecting growth in net income as well as the favorable impact of lower interest rates", call 2026-08-07 [3], Walsh), and the Corebridge deconsolidation moved a business out of the base. AIG's cleaner measures only exist for the recent period: adjusted tangible book value per share $67.62 → $70.37 over 2025 and core operating BVPS $61.75 → $69.12 (release 2026-02-10), both moving well.

Verdict: neutral. Not compounds: five years of book value per share going sideways, with the total return on book at 1.7% a year, is not compounding however good the recent quarters look. Not destroys either, and not close to it: the buybacks were done below book, retired a third of the company, cut leverage by 69%, lifted core operating ROE from 9.1% to 11.1%, and nothing has been written off. The fair reading is a company that spent five years converting a conglomerate into a smaller, better P&C insurer and handed the proceeds back at a good price — a restructuring, not a compounding record — and which now, for the first time in the window, has to fund its capital return out of underwriting. ⚠️ The kill criterion this implies is written into scorecard.md.

S3 addendum — earnings quality

Sources: the non-GAAP definitions and reconciliation in the release of 2026-02-10; XBRL companyfacts; the FY2026-Q2 call of 2026-08-07.

  1. The non-GAAP bridge. Flag 1 — restructuring has been a "one-off" for six years running. FY2025: net income to AIG common $3,096m ($5.43 diluted) → adjusted after-tax income $4,044m ($7.09 diluted), a 31% uplift; FY2024: a net loss of $(1,426)m against AATI of $3,254m (release 2026-02-10, Financial Summary). The release names twenty categories excluded from adjusted pre-tax income. Two of them are worth naming:
  2. "restructuring and other costs related to initiatives designed to reduce operating expenses, improve efficiency and simplify our organization" — RestructuringCharges $435m (2020), $433m (2021), $423m (2022), $356m (2023), $745m (2024), $439m (2025). Six years of six, and $2,396m across the five-year table above. A charge present in every year of six is an operating expense with a different label.
  3. "the portion of favorable or unfavorable prior year reserve development for which we have ceded the risk under retroactive reinsurance agreements and related changes in amortization of the deferred gain" — i.e. development on the ceded legacy book does not reach AATI. Recorded, not scored: it is a defensible exclusion and it matters for reading flag 4.
  4. Stock-based compensation. Clean. $207m in FY2025 = 0.77% of revenue and 6.2% of operating cash flow, against $5,836m of buybacks in the same year. It is falling, not rising ($278m in 2021). Nothing here.
  5. Cash conversion. Flag 2, and it is the one to watch. Operating cash flow: $6,243m (2023) → $3,273m (2024) → $3,314m (2025). Over the same stretch AATI went $3,254m → $4,044m (+24%) and AATI per diluted share $4.95 → $7.09 (+43%).
  6. OCF / net income to common: 1.71x (2023), not meaningful (2024, a loss), 1.07x (2025).
  7. OCF / AATI: 1.92x (2023) → 1.01x (2024) → 0.82x (2025). Adjusted profit now exceeds the cash the business generates. Part of that is structural and honest — Corebridge's cash flows left with the deconsolidation, and net premiums earned are roughly flat ($25,564m → $23,537m → $23,751m) — but the direction is unambiguous and it is the opposite of the earnings line. Accruals ((net income − OCF) / average total assets) are −0.14% in FY2025, essentially zero; the earlier years are not comparable because total assets fell from $539bn to $161bn on deconsolidation, so the ratio is reported for 2025 only.
  8. Sector equivalent — prior-year reserve development as a share of pre-tax income. Flag 3, and it lands on S8's row 3. XBRL SupplementalInformationForPropertyCasualtyInsuranceUnderwritersPriorYearClaimsAndClaimsAdjustmentExpense (negative = favourable) against income from continuing operations before tax:
FY2021FY2022FY2023FY2024FY2025
PYD ($m)+78 adverse(356)(234)(120)(427)
Pre-tax income ($m)13,3473,7722,8673,8703,879
Favourable PYD as % of pre-tax income−0.6%9.4%8.2%3.1%11.0%

Favourable development more than tripled in 2025 — $120m to $427m — in the year the 10%+ core operating ROE target came due, and it did so on a pre-tax income line that was otherwise flat ($3,870m → $3,879m). The arithmetic, stated with its caveat: the extra $307m of pre-tax release is about $242m after tax at 21%; had it not been there, AATI of $4,044m would be ~$3,802m and core operating ROE (on average core operating equity of $36,482m, release 2026-02-10) would be ≈10.4% instead of 11.1% — still above the target, with 40bp of headroom instead of 110bp. The caveat is that AATI excludes the retroactively-reinsured portion of development (see flag 1), so not all $427m reaches it; the calculation is an upper bound on the effect, not a restatement. It continues: H1 2026 underwriting income is up 68% on "lower catastrophe losses and more favorable prior year reserve development" (call 2026-08-07 [3], Walsh). And the analysts are circling it — Meyer Shields asked about IBNR in accident years '21 and '22 and was answered about '16 and '23 (§S8 dodge 3). This is the number to re-run on the Q3 2026 10-Q.

  1. Working capital / deferred costs. Mild note. Deferred policy acquisition costs $2,117m (2023) → $2,065m (2024) → $2,106m (2025) against net premiums earned of $25,564m → $23,537m → $23,751m: DAC/NPE 8.3% → 8.8% → 8.9%. Rising two periods running, but on roughly flat earned premium, so the rule's "while revenue grows" condition is not met; it is also consistent with the acquisition-ratio increase management attributes to mix (call 2026-08-07 [17], Walsh). Mild, watched, not a finding. No separately tagged capitalised-software or contract-cost line: recorded as not disclosed rather than clean.

EQ summary: 3 flags — restructuring charges excluded from the headline in six consecutive years ($2,396m over five); operating cash flow at $3.3bn against $4.0bn of adjusted after-tax income, i.e. cash conversion below 1.0x and falling while adjusted EPS rose 43%; and favourable prior-year development tripling to 11.0% of pre-tax income in the year the ROE target came due. One mild note on DAC. Clean: stock-based compensation, and accruals at −0.1% of assets.

Full notes, 23 Sep 2026 13 min read

AIG — dossier, 2026-09-23

Slot names-insurance-2, second member. Model-drafted unattended by the cloud research routine. Nothing here is human-verified. Every figure carries the document it was read from; anything from background knowledge is marked [background]. Quotes were located by exact substring search in a text extraction of the filing and copied out — never re-typed.

Stage 0 for this name is _clusters/cluster-insurance-rescore-2026-09-22.md §5 (lens 3/4). This dossier does not re-score it; it records evidence that contradicts two of its inputs.


1. Two Item 5.02 filings, fifteen days apart

FiledItemContent
2026-09-02 (event 2026-09-01)5.02Peter Zaffino notified the Board that "he will step down as Executive Chair and a member of the Board, effective September 15, 2026", becoming Senior Advisor to the CEO. "John Rice, the Company's Lead Independent Director, will become Chair of the Board". The filing states his "resignation is not due to any disagreement with the Company, its management or the Board", and an amendment to his November 2022 employment agreement confirms no compensation change and that the transition is neither a termination without Cause nor one for Good Reason.
2026-09-165.02Jon Hancock, EVP and CEO, General Insurance, "will retire and transition to the role of Senior Advisor, effective December 31, 2026", after more than six years in executive leadership at AIG and a 40-year career. He will report to Eric Andersen, President & CEO.

Why this is the run's finding rather than a governance footnote. After the Corebridge exit completed in May 2026, General Insurance is AIG. Its CEO is leaving. The Executive Chair who ran the transformation left the Board two weeks earlier. And AIG's own FY2025 10-K carries, in the Item 1A Risk Factor Summary on p.12:

"Competition for employees in our industry is intense, and managing key employee succession is critical to our success."

Hancock was on the Q2 call on 2026-08-07 — the CEO's prepared remarks introduce him as joining for Q&A as a sitting segment head. Forty days later he is retiring.

The market did not react. ALL's shares moved −9% in three sessions after its August catastrophe release; AIG closed $75.88 on 2026-09-01 and $76.07 on 2026-09-16 — the two announcement days — and $75.18 on 2026-09-22 (market data NYSE:AIG). Volume on 2026-09-18 was 7.7m against a 21-session average of about 4.1m, but the price did not move with it. This dossier records that as an observation, not as evidence of anything.


2. Two inputs to the Stage 0 zero have reversed

The re-score gave AIG 0 on test 4, "growth that does not dilute", on this reasoning: buybacks of $6,652m (2024) and $5,836m (2025) against operating cash flow of $3,273m and $3,314m, with net earned premium down 24% and reserves down 11% over five years — capital return funded by divestiture rather than underwriting. H1 2026 says the opposite on both legs.

Capital return is now inside cash flow. From the Q2 2026 10-Q cash flow statement:

Six months ended 30 June ($m)20262025
Net cash provided by operating activities1,8711,335
Purchase of common stock(1,153)(4,007)
Dividends on common stock(504)(488)
Issuance of long-term debt—1,241

Buybacks fell 71% year on year and now sit below operating cash flow even after the dividend. $2.6bn remained authorised at 2026-07-31, and "Pursuant to a Rule 10b5-1 plan, from July 1, 2026 to July 31, 2026, we repurchased approximately 2 million shares of AIG Common Stock for an aggregate purchase price of approximately $ 195 million." — an annualised pace of roughly $2.3bn, less than half the 2025 rate.

Premiums are growing. From the Q2 2026 10-Q General Insurance table:

($m)Q2 2026Q2 2025ChgH1 2026H1 2025Chg
Net premiums written7,5166,880+9%13,11511,406+15%
Net premiums earned6,1965,878+5%12,24811,647+5%
Underwriting income686626+10%1,460869+68%
Accident year combined ratio, as adjusted88.188.4(0.3)87.388.2(0.9)
Catastrophe losses and reinstatement premiums (pts)(3.4)(2.9)(3.2)(6.0)
Prior year development, net (pts)2.52.02.41.6

The five-year shrink the cluster file used ends at 2025. A re-score belongs in a dated cluster file, not in a name pass (ledger/research/README.md: never edit a cluster file after its date), so this is recorded as evidence for one rather than acted on here. The scorecard's S0 line carries a pointer to this section.


3. What the growth is, and what it costs

From the Q2 call (2026-08-07), CEO prepared remarks and CFO review:

  • Net premiums written +9%, or +11% excluding North American property. Global commercial +9%, North America commercial +9%, international commercial +10%, global personal +8%.
  • Global commercial retention 88%; new business $1.9bn, +37% year on year, including the strategic transactions.
  • Lexington property premium retention fell nine percentage points in the quarter, deliberately: management says it is contracting the portfolio where E&S property pricing is inadequate, and that this and the pricing environment together reduced overall North America growth by over three percentage points.
  • Pricing, Q2 2026: North America commercial ex-property +5%; North America retail casualty +10% with excess casualty +14%; financial lines ex-cyber flat; international commercial renewal pricing −6%, with global energy −15%. Political violence and terrorism +9% after −8% in Q1.
  • Capital returned in Q2: $904m — $641m buybacks, $263m dividends.
  • Expense ratio: Q2 30.8%, trailing twelve months 30.7%, target below 30% for full-year 2027.
  • Net investment income $871m in General Insurance, flat year on year; core fixed income +4% with new-money yields about 60bp above sales and maturities, annualised yield 4.72%; alternatives fell to $13m from $48m, private equity a $8m loss on a one-quarter lag.
  • Corebridge is fully exited: ~25m shares sold for $710m in May 2026, completing the five-year separation.
  • Debt $9bn, total debt to adjusted capital 17.6%. BVPS $77.39 (+4% y/y); adjusted tangible BVPS $72.18 (+3%).

On the buyback, the CEO's stated logic is explicitly a book-value argument:

"Given our current share price at a modest premium to tangible book value, we view the repurchase of our shares as a very attractive use of capital." — Q2 FY2026 call, CEO prepared remarks

At $75.18 against $72.18 of adjusted tangible book, that premium is about 4%.


4. The reserve composition, and a direct contradiction with ALL

Q2 2026 prior-year development, from the CFO's prepared remarks: $145m favourable net, comprising $146m of net favourable loss reserve development, $26m of ADC amortisation and $27m of prior-year return premiums. The composition:

LineQ2 2026
U.S. workers' compensation+$177m favourable
U.S. property and special risks+$79m favourable
U.S. excess casualty−$74m adverse, "predominantly in accident years 2016 and 2023"

This is the same shape the ACGL pass found on 2026-09-22 — short-tail and run-off lines releasing while the long tail strengthens — in a different issuer, and it is disclosed voluntarily here rather than extracted from a table.

And then the contradiction. Asked what he is seeing on social inflation, AIG's CEO said:

"we haven't seen anything that says that it's moderating at this point. And we're certainly not building that into our pricing at this stage." — Q2 FY2026 call, answer to Mayor Shields (KBW)

Allstate's CEO, on a call the day before (2026-08-06), explained four consecutive quarters of large favourable auto development by pointing at tort reform in Florida, Georgia and Louisiana and new legislation in New York, and said he is hopeful that trend continues (see ALL/dossier-2026-09-23.md §4). Two US carriers, one day apart, reading the same litigation environment in opposite directions — and only one of them is booking reserve releases on it. Personal auto and large-account commercial are not the same exposure, so this is not a straight contradiction of fact; it is a difference in what each management is willing to capitalise, and it is worth an owner's attention because the board's own scuttlebutt row recommends one of them.


5. The euro note issue

On 2026-09-16 AIG priced €1,125,000,000 of senior unsecured notes off shelf File No. 333-277075: €625m of 4.250% Notes due 2031 and €500m of 4.750% Notes due 2036, interest accruing from 2026-09-24, with application made to list on Euronext Dublin's Global Exchange Market. Net proceeds approximately €1,114m.

"We intend to use the net proceeds from this offering for general corporate purposes, which may include the repayment of the approximately €821 million outstanding aggregate principal amount of our 1.875% Notes due June 2027" — 424B2 filed 2026-09-16, "Use of Proceeds" summary, p. S-3

Read plainly: €821m of 1.875% money is being replaced with 4.250%–4.750% money — a coupon roughly 2.4 to 2.6 times higher on the refinanced portion — and about €300m of incremental debt is being raised on top. Against $9bn of total debt at 17.6% of adjusted capital that is not a solvency event; it is a cost-of-capital fact that belongs in any 2027-28 earnings model, and it is why one of the replacement kill criteria is about borrowing to buy stock.


6. Kill test (S2) — the numbers

FY figures from XBRL company-concept series (10-K accession 0000005272-26-000023); H1 2026 from the Q2 10-Q.

20212022202320242025H1 2026
Net earned premium ($m)31,28526,76525,56423,53723,75112,248 (GI)
Net income ($m)n/rn/rn/r(1,404)3,0961,742
Diluted shares (m)n/rn/rn/r657.3570.3537.8
Operating cash flow ($m)6,2234,1346,2433,2733,3141,871
Buybacks ($m)2,5925,2002,9616,6525,8361,153
Reserves ($m)79,02675,16770,39369,16870,666n/r
Equity ($m)65,95640,97045,35142,52141,13940,606

Equity has fallen every year since 2023 while shares outstanding fell 34% — the per-share arithmetic is the denominator, which is exactly what the re-score said. What is new is that the numerator turned up in 2026 and the denominator work slowed.

Diluted EPS H1 2026 $3.18 against $3.13 — flat, on an 8.6% smaller share count, so net income to common fell about 8%. Adjusted after-tax income per diluted share was $2.00 in Q2, +10% y/y; core operating ROE 11.1% in the quarter and 11.6% for the half. The gap between the GAAP and adjusted series is why S8 is modelled on the adjusted one and says so.


7. Ownership (S5)

3spread, 13F report date 2026-06-30 (threespread.py holders AIG; 764 filers against 1,108 — the quarter is incompletely ingested, no exit may be read from an absence; the named non-filers include both Vanguard entities, Geode, Franklin, GQG, Morgan Stanley, Capital World, Capital Research, Dimensional and Northern Trust).

ManagerShares$mQoQ
BlackRock48,935,9333,647.2+2,264,892
Wellington30,331,7042,260.6+1,961,270
State Street25,115,4161,871.9+196,969
Harris Associates20,357,2721,517.2−239,706
T. Rowe Price13,777,3171,026.8−97,989
Invesco13,725,7501,023.0+863,345
Hotchkis & Wiley11,841,030882.5+1,610,435
Diamond Hill6,232,910464.5−2,343,751
Ameriprise4,651,581347.0−6,731,349

Who is on the other side: index money plus a deep-value active roster (Harris, Hotchkis, Diamond Hill, LSV, Aristotle) that is a natural holder of a sub-book insurer, with Wellington adding.

The Sixth Street row is not a holding. 4,343,856 shares reported as NEW, byte-identical to Charles Schwab Investment Management's position — the defect documented in ledger/research/README.md, now seen at a sixth issuer after QCOM, ACGL, KNSL, AX and ALL.

Insiders: 22 current directors and executive officers hold 3,238,630 shares in total, and "None of the directors, the named executives or the directors and executive officers together as a group owned more than one percent of our common stock" (DEF 14A 2026-03-31). On 524.7m shares outstanding that is about 0.6% — against Allstate's 1.55% in the same cluster. Zaffino's 218,463 options and RSUs exercisable within 60 days are inside that total, and he has now left the Board.

Short interest (Equibles/FINRA): 8,709,587 shares at the 2026-08-31 settlement, up from 7,423,123 at 2026-07-31, 3.2 days to cover — about 1.7% of shares outstanding. Negligible, and rising slowly. (Equibles also returns a model estimate for the 2026-09-15 settlement of ~9.24m shares; that is a prediction, not a FINRA figure, and is not used above.)


8. Filings read, and the red-flag sweep (S3)

Three biggest risks, in management's own ordering, from the FY2025 10-K Item 1A Risk Factor Summary (p.11), which is explicitly "a summary of the material risks and uncertainties":

  1. Market Conditions — "Deterioration of economic conditions, geopolitical tensions, changes in market conditions or weakening global capital markets have affected and may continue to materially affect our businesses, results of operations, financial condition and liquidity."
  2. Reserves and Exposures, first bullet — "The amount and timing of insurance liability claims are difficult to predict and such claims may exceed the related liability for unpaid losses and loss adjustment expenses."
  3. Reserves and Exposures, second bullet — "Reinsurance may be unavailable or too expensive relative to its benefit and may not be adequate to protect us against losses."

Auditor: PricewaterhouseCoopers LLP, PCAOB ID 238. Sole critical audit matter: valuation of loss reserves, net liability $41.8bn at 2025-12-31, with procedures including controls "designed to identify and address management bias and contrary evidence" and an independent PwC reserve estimate for a sample of product lines.

Red-flag sweep (efts.sec.gov full text, CIK 0000005272, 2023-01-01 to 2026-09-23):

QueryHitsWhat they are
"changes in accountants"0—
"non-reliance"1an exhibit to the Q3 2024 10-Q, not a non-reliance determination
"material weakness"610-K and annual-report ICFR boilerplate
"restatement"27clawback-policy and award-agreement exhibits

No NT 10-K; no 10-K/A or 10-Q/A. The only amendments in the index are 13F-HR/A, 3/A, 4/A, 8-K/A and 13G/A. Clean.


9. Scuttlebutt (S6) — and a staleness warning

Carried from _clusters/cluster-insurance-2026-09-22.md §4, two claim rows from the 2026-09-07 Real Eisman Playbook: "I am recommending AIG" (Ryan Tunis) and "And full disclosure, everybody, I own AIG" (Steve Eisman, described there as long and recently added). Those quotes live in the board's claim files.

Both were recorded on 2026-09-07 — six days after Zaffino's resignation was filed and nine days before Hancock's retirement was. Neither speaker is reacting to either. This is the second time in two slots that carried-forward S6 has gone stale inside a fortnight (QCOM's did on 2026-09-22), and it is now a pattern worth an owner decision: cluster scuttlebutt has a shelf life of roughly two weeks on any name where Item 5.02 or a Reg FD 8-K can fire.

No new outside evidence was gathered for AIG this run — the S6 search budget went to ALL. The product test and the expert call stay ⏳ OPEN (user).


10. Evidence that is thin, and access limits

  • One transcript, not two. Only the Q2 FY2026 call was read, and only turns 3–16 of 52 — the prepared remarks plus the first two Q&A exchanges. The rest of the Q&A, including anything said about succession, was not reached.
  • No new S6 for this name. The only scuttlebutt is the carried-forward cluster row, which this dossier flags as stale. No analyst-coverage count and no price targets were gathered for AIG.
  • No investor supplement and no Investor Day deck, though management sources its ROE and expense targets to the 2025 Investor Day. The Everest Colombia acquisition and the Everest renewal-rights transaction exist here only in management's spoken remarks, in no filing this run read.
  • Form 4 transaction codes not itemised, so no insider buy/sell direction is asserted. In particular, nothing is claimed about whether Zaffino or Hancock have sold.
  • Next-earnings date is derived, not scheduled — from the 2024-11-04 and 2025-11-04 Q3 results 8-Ks. Equibles' IR calendar was not queried for this name (budget), and market data's needs a paid plan.
  • The €821m/1.875% refinancing is stated as an intention in the prospectus ("may include"), not a completed redemption.
  • Transcript speaker labels are unreliable — analyst names appear on management turns and the first questioner's name is wrong. Nothing was attributed from this source without an internal check inside the turn itself. The 8-K spelling "Andersen"/"Jon Hancock" is preferred over the transcript's "Anderson"/"John Hancock" throughout.