Tomorrow’s Publisher — source and forecast notes Bloomsbury’s smaller AI cheque puts recurring income to the test 23 October 2026 • Ivan Massow Forecasts are Noah modelled judgments, not observed frequencies. No matched empirical reference-class denominator was supplied. Uncertainty intervals below are the returned model ranges; they are not an assurance of calibration. Dates and odds have not been altered. All five returned estimates are used; two cover the same event at different dates. Four concern public disclosure, not the existence of undisclosed income or contracts. A later tested date is not an inferred exact crossing date. 1. Will Bloomsbury state in an RNS, annual report or results release published by 2027-05-31 that it executed after 2026-10-23 a new, renewed or extended paid licence allowing an AI company to use Bloomsbury content for model training, retrieval or generated outputs? YES: 46.75% | NO: 53.25% Returned YES uncertainty range: 25–68%. Deadline: 2027-05-31 YES condition: YES if, by 2027-05-31, Bloomsbury has published an RNS, annual report or results release stating that it executed after 2026-10-23 at least one qualifying new, renewed or extended paid AI-content licence. NO condition: NO if, by 2027-05-31, Bloomsbury has not published a qualifying statement about such an agreement executed after 2026-10-23; this does not establish that no undisclosed agreement exists. Meaning: The forecast is a publication event: it settles on Bloomsbury disclosing a qualifying executed licence. NO means no qualifying disclosure by the deadline, not proven nonoccurrence of an undisclosed deal. Origin: Noah final quick forecast, authenticated final.v2 reassessment. Run reference: acc8ced3fb0048a4212efeacbcbb6fc11db33f74807bf08fdcf46ef740f921d3 Final receipt digest: dbf3295464915af0d70aefb58dfeb06b7fa9d46fb9808ed6fe75bde2305d3502 2. Will Bloomsbury state in an RNS, annual report or results release published by 2027-10-23 that it executed after 2026-10-23 a new, renewed or extended paid licence allowing an AI company to use Bloomsbury content for model training, retrieval or generated outputs? YES: 52% | NO: 48% Returned YES uncertainty range: 30–72%. Deadline: 2027-10-23 YES condition: YES if, by 2027-10-23, Bloomsbury has published an RNS, annual report or results release stating that it executed after 2026-10-23 at least one qualifying new, renewed or extended paid AI-content licence. NO condition: NO if, by 2027-10-23, Bloomsbury has not published a qualifying statement about such an agreement executed after 2026-10-23; this does not establish that no undisclosed agreement exists. Meaning: The forecast is a publication event: it settles on Bloomsbury disclosing a qualifying executed licence. NO means no qualifying disclosure by the deadline, not proven nonoccurrence of an undisclosed deal. Origin: Noah final quick forecast, authenticated final.v2 reassessment. Run reference: 287b1c39f2dca37ceb0f5536d41fcc778038dce40dfa772dec588e3309817152 Final receipt digest: cd071869ea0670866253f98922045b0e58de48b343d4dcfd1c9fe0d93322e46c 3. Will Bloomsbury publish by 2027-06-30 a standalone sterling amount for AI-content licensing income earned in its financial year ending February 2027, separate from other Academic & Professional revenue? YES: 25% | NO: 75% Returned YES uncertainty range: 12–42%. Deadline: 2027-06-30 YES condition: YES if, by 2027-06-30, a Bloomsbury results release or annual report gives a standalone sterling amount for AI-content licensing income earned in the financial year ending February 2027. NO condition: NO if, by 2027-06-30, Bloomsbury has not published such a standalone amount, including if it describes AI licensing only qualitatively or embeds it in a broader revenue figure. Meaning: The forecast concerns publication of a separate income figure, not whether Bloomsbury earned undisclosed AI-licensing income. Origin: Noah final quick forecast, authenticated final.v2 reassessment. Run reference: 82b2bcffa503956d2e6bbc5ea3d17dc7701f175f98464d2a9aed026eb14c55a6 Final receipt digest: d17dc83a05a424294125f8c48b7801c94dbcab159fb38a9044667a82bed3121e 4. Will Bloomsbury's Academic & Professional adjusted operating margin for the financial year ending February 2027 be at least 18%, as calculable from its full-year public financial records available at 2027-10-23? YES: 43% | NO: 57% Returned YES uncertainty range: 24–63%. Deadline: 2027-10-23 YES condition: YES if Bloomsbury's full-year public financial records available at 2027-10-23 show Academic & Professional adjusted operating profit equal to at least 18% of divisional revenue for the financial year ending February 2027. NO condition: NO if Bloomsbury's full-year public financial records available at 2027-10-23 show that margin below 18%. Meaning: The event is the division's full-year economic result; Bloomsbury's results and annual report verify it rather than constitute it. Origin: Noah final quick forecast, authenticated final.v2 reassessment. Run reference: 0708b1cc4983775ac5104a70f4101a342f8696c5ddfce726b0e3551bb5043ef3 Final receipt digest: 146394edc9709639f0a7b7e5d44ddc31e7288b17bbb57d76f6201c81a276934d 5. Will at least two UK-listed publishing groups other than Bloomsbury publish by 2027-10-23 that they executed paid agreements licensing owned or controlled text corpora to AI companies for model training, retrieval or generated outputs after 2026-10-23? YES: 52% | NO: 48% Returned YES uncertainty range: 30–72%. Deadline: 2027-10-23 YES condition: YES if, by 2027-10-23, qualifying RNS releases, annual reports or official company statements from at least two distinct UK-listed publishing groups other than Bloomsbury say they executed such paid agreements after 2026-10-23. NO condition: NO if, by 2027-10-23, fewer than two distinct qualifying publishing groups have published such statements; undisclosed agreements do not qualify. Meaning: The forecast is explicitly about public disclosure by peer publishers, because visible transactions are what create editorial and negotiating benchmarks; NO does not rule out confidential deals. Origin: Noah final quick forecast, authenticated final.v2 reassessment. Run reference: c91dd303ca3bbd5124ad5f7cb778742e417f8c9f8feca2ed2c6d71c555b96bac Final receipt digest: 7395c80f85db79563a2982797d3e91192920dee05be46cf463cef88e15237083 Why these forecasts matter Repeat licensing: prior same-company execution and non-exclusive rights make another agreement plausible; neither expiry nor fresh negotiations have been verified. The execution and publication conditions both matter. Disclosure: existing qualitative and bundled reporting supports a lower chance of a separate income figure. It says nothing about whether income exists. Margin: the full-year outcome depends on the whole division’s revenue and costs, including second-half mix and licence contribution. Consumer-driven group guidance is not academic-division guidance. A stronger second half is possible, but not the more likely central forecast for the exact18% threshold. Peer deals: established licensing and demand support possible further transactions, but two distinct UK-listed publishing groups must each execute and disclose a qualifying paid licence. US-listed Wiley is not a qualifying group in that forecast. Fact checks and limits Bloomsbury’s interim period ends31August2026; the release is dated22October2026. Academic revenue48.8m versus46.1m; adjusted profit6.8m versus11.0m. Management says the continuing licence contributed a lesser quantum and links this to reduced margin. The full causal profit bridge and AI-specific income are not disclosed. Do not infer licence price cuts, a lost buyer or an expired licence. The financial statements note3 show Academic & Professional adjusted operating profit7.0m on48.8m revenue:14.3443%, rendered about14.3% in the article. Headline adjusted profit6.8m includes finance and gives13.9%; these are different measures. The18% forecast is the operating measure. A server-supplied contextual angle confused these figures; the host identified the difference before final calculation and the final rationale uses the corrected operating basis. No target, date or odds were edited. Group confidence in record full-year results remains in the article. One sentence in the saved final model rationale referred to higher group half-year revenue/profit; the actual filing reports lower group H1 figures. That sentence was excluded from editorial reasoning. It does not change the stored authenticated odds, and is flagged in coder feedback. The£1m/£100,000/£400,000/£300,000 example is expressly hypothetical. It describes no company’s undisclosed payment schedule. Informa figures are Taylor & Francis divisional underlying growth, not group growth. Do not calculate an undisclosedAIfee by subtracting growth rates or compare margins with Bloomsbury. Wiley figures are company-reported in its2026 shareholder letter: recurringAIrevenue1mto8m, totalAI49m and19corporateAI-subscription knowledge-feedcustomers. Definitions and periods differ from UK publishers; no relative profitability ranking is made. Training/retrieval descriptions explain mechanisms. Recommendations about budgets and commercial checks are editorial judgment, not probability forecasts or legal advice. Research limits The initial Noah shared programme completed its two required components and selected5native automatic forecast targets. All5received final estimates after a blind host contribution. Nine targeted initial hypothesis tests returned no admitted evidence; this does not prove absence or corroborate the hypotheses. Broad lexical-match counts and stock/SEO/adjacent content were not used as measures of commercial adoption or profit. No bounded retrieved sample is presented as a market denominator. Fresh primary accounts provide the actual financial findings. Source snapshots and full returned research/forecast JSON are preserved privately. Research identity Programme: nabdl_programme_run_94f2659892ffdcd41b1ec86838edd59c2462b940affa6afa Jobs: nabdl_job_a96c8b3a5e3ee89fdbda0a9a8491147932fd25058d994d95bf2772d436f56bb5 nabdl_job_e43241b8f12cac59edc8e86941cefdce4967a2c0e29cd1ecdad2faf0ed225336 Post-draft check: nabdl_job_04b3a1592caffa075b4cedfb98f9c65c43fa6af2408569e9b5e22e42d1525b7c Sources Bloomsbury: interim results, 22 October 2026 https://www.bloomsbury-ir.co.uk/media/press_releases/2026/221026.asp Bloomsbury: interim financial statements, including segment operating profit https://www.bloomsbury-ir.co.uk/docs/librariesprovider16/archives/annual_reports/interim26.pdf Bloomsbury: full-year results and author opt-in, 20 May 2026 https://www.bloomsbury-ir.co.uk/media/press_releases/2026/200526.asp Informa: 2026 half-year results https://www.informa.com/media/press-releases-news/latest-news/informa-2026-half-year-results/ Wiley: 2026 shareholder letter and recurring AI revenue https://www.sec.gov/Archives/edgar/data/107140/000010714026000041/jwa-20260813.htm Post-draft investigation: all six support/falsification lanes completed, but returned no admitted evidence for the three defined commercial propositions. Broad software, music and generic content-marketing results were not treated as publisher renewal evidence. No extra forecasts were commissioned.