Gap Scan — 2026-05-26

Live run. Six new pages were researched and committed directly to their real Wiki/ folders (status: active; all confidence: medium for new pages, confidence: high for Peter Drucker as an uncontested historical record). One thin anchor was expanded. Two foundational primary sources were authenticity-verified and queued for ingest — closing the Pope Leo encyclical deferral that the 2026-05-25 gap-scan flagged, and capturing the Topaz/Lancet AI-fabricated-citations paper that today's dev-log surfaced. No authenticity-verification failures.

Scanned

Recent window (last 48h): 3 New Developments Log/ files (2026-05-25-0511, 2026-05-25-1511, 2026-05-26-0511); ~62 wiki pages with last_updated inside the window — heavily concentrated in the 2026-05-25 gap-scan's live-build cluster (Knight Columbia author entities, ILAI thin-anchor expansions, the concepts/classification-institutions build, and the Pope Leo / Anthropic / Matthew Prince / Cloudflare threads).

Rotation slice: 5 of 14 — entities/ H–N (~95 pages).

Broken-link analysis: 559 unique broken targets across the wiki (down only modestly from yesterday's 556; the 2026-05-25 run cleared 13 alias clusters but the day's edits surfaced new ones). After alias resolution, doc-artifact filtering, and dropping items already actioned in the 2026-05-22 / -23 / -23-run2 / -24 / -25 gap-scan reports — and items with an existing Raw Sources/ file or queued INGEST- task — ~25 genuine new gaps remained.

The dominant cluster today: the Pope Leo encyclical (now stable; Raw Sources/ PDF in place since 2026-05-25 17:28 UTC), and the Cloudflare-Prince / ClickUp / Bosworth labor-displacement thread, which named four concept anchors with no pages behind them (Middle-Manager Displacement, AI Displacement vs. Augmentation, AI Fluency Divide, plus the Peter Drucker foundation). The rotation slice surfaced two thin anchors worth a same-run expansion (Institute for Law & AI at 10 inbound × 81 words; Henry Farrell at 17 inbound × 274 words, deferred to next run as the heavier expansion).

Gaps actioned (9 of ~25 found)

New pages created (live)

• Peter Drucker — gap type 3: foundational management-theory figure (1909–2005) explicitly cited from Matthew Prince and How I Choose Which Cloudflare Employees to Replace With AI as the intellectual root of the builders/sellers/measurers framework; lint-report-flagged (2026-05-22) and deferred since. Created at confidence: high — uncontested historical record. 3 inbound, plus a fresh pair of inbound from the new Middle-Manager Displacement and AI Displacement vs. Augmentation pages.

• Middle-Manager Displacement — gap type 2 + live thread: named in How I Choose Which Cloudflare Employees to Replace With AI, Cloudflare, and Matthew Prince, with new live evidence today (ClickUp 22% layoff May 22; Bosworth Meta reorganization May 24; Gartner survey May 5 on AI-driven cuts). Built on the Drucker typology + 2026 layoff pattern. 3+ inbound.

• AI Displacement vs. Augmentation — gap type 2: the prior analytic question presupposed by middle-manager-displacement and by AI Labor Disruption generally; named in How I Choose Which Cloudflare Employees to Replace With AI and now reinforced by the WSJ "AI Natives" piece (2026-05-25), Anthropic Economic Index Q1 2026, and the Lancet endoscopist-deskilling case. 4 sources; balanced "displacement vs. augmentation" matrix across occupation clusters.

• AI Fluency Divide — gap type 2 + live thread: named in UC Berkeley School of Law AI Policy (Effective Summer 2026) and central to today's WSJ "AI Natives" framing (22% of 18–24-year-old US degree-holders feel "very prepared" — highest of any age group). Built as the third leg of the Drucker / displacement-vs-augmentation / fluency-divide labor-policy triad. 3 sources.

• Dominion Energy — gap type 3 + live merger thread: 3 inbound (from lint-report, How the AI revolution has turbocharged M&A, Inside the Dirty, Dystopian World of AI Data Centers — Matteo Wong (The Atlantic, April 2026)) and central to the May 18, 2026 NextEra acquisition that the wiki tracks on AI Data Centers. Built with the canonical merger terms (0.8138 exchange ratio, $67B all-stock, ~240 GW combined footprint) and the standing 5.5%/year Virginia demand-growth anchor.

• NextEra Energy — gap type 3 + live merger thread: 2 inbound and the acquirer in the May 18 deal. Built as the renewable-generation counterpart to Dominion's transmission/distribution footprint; flags the hyperscaler-PPA structural role (Amazon, Google, Microsoft, Meta).

Schema flag. Both new utility pages (entities/dominion-energy, entities/nextera) are for-profit companies, not non-profit / non-company organizations. CLAUDE.md's "Where does X go?" rules place for-profit AI developers in companies/ and everything else in entities/. Utilities are neither — they are for-profit but not AI developers. I placed both under entities/ consistent with existing wiki pattern for utility actors that recur across AI-data-center sources, but the categorization deserves a CLAUDE.md schema-pass clarification at next opportunity. Flagged inline on both pages.

Pages expanded (live)

• Institute for Law & AI — thin anchor: in-degree 10, only 81 words. Expanded to ~400 words. Folded in the Cambridge anchoring, Christoph Winter's director designation (corroborating yesterday's Christoph Winter factual correction), Bullock's Senior Research Fellow designation (likewise), the institute's EU GPAI Code of Practice advisory engagement, and the Radical Optionality (Winter & Bullock, 2026) flagship-essay context. 1 supporting cite (law-ai.org); held at confidence: medium. Closes the ILAI thread the 2026-05-25 run opened with the Winter and Bullock expansions.

Link aliases fixed

• None this run. The 2026-05-22 / -23 / -24 / -25 cumulative passes have absorbed the standing alias backlog (AISI, EFF, NVIDIA, interpretability, jailbreaking, x-risk, california-AB-2013). The single remaining concepts/jailbreaking instance flagged by today's scan is in Wiki/log.md as historical record ("...(planned)..."); the gap-identifier protocol does not edit log.md, and the substantive references have all been migrated.

Queued — foundational sources

• Pope Leo XIV's Magnifica Humanitas encyclical — Pope Leo XIV's first encyclical and the Catholic Church's first sustained magisterial document on AI; ~42,300 words; promulgated 2026-05-25 (signed 2026-05-15), unveiled at the Vatican's Synod Hall with Anthropic co-founder Chris Olah present. Verified: canonical host vatican.va, full canonical URL https://www.vatican.va/content/leo-xiv/en/encyclicals/documents/20260515-magnifica-humanitas.html, corroborated by NYT, NPR, TechCrunch coverage and the vatican.va promulgation-event page; PDF (51 pages, 1.5 MB) downloaded 2026-05-25 17:28 UTC; structure and distinctive-passage check passed (chapter-tree on page 1 matches reportage). Saved: Raw Sources/Encyclical Letter of His Holiness Leo XIV Magnifica Humanitas (15 May 2026).pdf (was already in Raw Sources/ pending stability check from the 2026-05-25 deferral; now cleared). Queued: Wiki/queue/INGEST-magnifica-humanitas-leo-xiv-2026.md. Verification record: Magnifica Humanitas Leo Xiv 2026. Closes the same-day-of-publication deferral the 2026-05-25 gap-scan flagged.

• Topaz et al., Fabricated citations: an audit across 2·5 million biomedical papers — peer-reviewed Lancet correspondence (vol. 407 issue 10541, pp. 1779–1781; DOI 10.1016/S0140-6736(26)00603-3) documenting a 12-fold rise in AI-fabricated references in PubMed-indexed papers (1 in 2,828 in 2023 → 1 in 277 in early 2026; 98.4% not retracted). Surfaced by today's dev-log (Fortune coverage 2026-05-24). Verified: canonical host thelancet.com, DOI resolvable, byline cross-checked against Columbia Nursing press release, Nature, Retraction Watch. Saved: to be pulled at ingest time (paywalled JS-rendered fulltext page is best fetched fresh; verified canonical URL recorded). Queued: Wiki/queue/INGEST-topaz-fabricated-citations-lancet-2026.md. Verification record: Topaz Fabricated Citations Lancet 2026. Will become the wiki's primary numeric anchor for the "scientific papers" leg of AI Content Saturation (\"AI Slop\") and supplements Sycophancy and Hallucination.

Authenticity-verification failures

None. Both queued sources verified cleanly against canonical hosts (vatican.va, thelancet.com), with byline + identifier + distinctive-passage checks all passing.

Notes for the next lint / ingest pass

• Schema clarification needed. Utilities (Dominion, NextEra) are for-profit but not AI developers; current CLAUDE.md "Where does X go?" rules do not cleanly cover them. Both placed under entities/ consistent with wiki precedent; schema pass should clarify. Same question pending for the broader Kkr / Blackstone / Blackrock / Apollo finance-firm cluster from How the AI revolution has turbocharged M&A (still in deferred backlog).
• entities/dominion-energy and entities/nextera frontmatter. I used entity_type: nonprofit as a placeholder for both because the existing schema's allowed values (person | nonprofit | think-tank | standards-body | coalition | pac | regulator | intergov | religious-body | academic-institution) do not include utility or for-profit-non-AI. Flagged inline on both pages and here for resolution at the next CLAUDE.md schema pass; recommend adding an entity_type: for-profit-other or similar value.
• Anthropic-Pope partnership thread. With the encyclical queued, the wiki has a clear path to upgrade citations across Pope Leo XIV, Chris Olah, Anthropic, and Religious and Civil-Society Voices on AI from (Source: news URL) to [[sources/magnifica-humanitas-leo-xiv-2026]] once the ingest cycle runs.
• Maxim Topaz entity page. Becomes a 2+ inbound entity after the Lancet ingest runs (the dev-log already mentions him by name); good candidate for the next gap-scan run.
• Andrew Bosworth, Allison Shrivastava, Paolo Carozza, Maxim Topaz, Zeb Evans, Ben Broca — all named in today's two new dev-log files. None are urgent (none are >2 inbound yet), but the Bosworth / Meta-AI-first thread is rising fast and Carozza (Pontifical Academy of Social Sciences / Meta Oversight Board, per TechCrunch 2026-05-25) is the next natural Pope-encyclical companion entity.

Deferred backlog (over the daily cap — re-surfaces next run)

• Henry Farrell — thin anchor in rotation slice: in-degree 17, 274 words, confidence: medium, sources_count: 1. Highest-priority next-run expansion. Score 5.
• Maxim Topaz — author of the queued Lancet paper; will be 2+ inbound after ingest. Score 3.
• Claude Code — broken-link cluster, 5 inbound (across Claude Mythos Preview, Stainless, KPMG, Andrej Karpathy). Quality-gate-ambiguous (concept page vs. product-anchored sub-section of Anthropic); needs a scoping decision. Score 4.
• Paolo G. Carozza — Notre Dame Law, Pontifical Academy of Social Sciences, Meta Oversight Board chair (per TechCrunch 2026-05-25); natural follow-on to the encyclical ingest. Score 3.
• Andrew Bosworth — Meta CTO driving the "AI-first" reorganization (per WSJ 2026-05-24); rising live thread. Score 3.
• Carried from 2026-05-25: Victor Zhenyi Wang (score 3); AI and Children (score 3); John Moolenaar (score 3); Seeing Like a State (applied to AI) (score 3); AI Political Economy (score 2); Center for a New American Security (CNAS), National Security Commission on Artificial Intelligence (NSCAI) (score 2 each); Pim de Witte (score 2); Ai Coarse Grainings (score 2); Autonomy Certificates (score 2); Ai Risk Management (score 2); Chinese AI Policy (score 1); Fifth Amendment And Ai (score 1); Jack Dorsey, Brian Armstrong (score 1 each); Mulligan-cluster stubs (score 1 each).
• Carried lint cleanups (not gap-identifier pulls): sources/anthropic-pwc-expanded-partnership (4 broken — likely supporting, folds into Anthropic); sources/trustworthy-agents-in-practice (4 broken — log.md records the page as both created and deleted). Flagged for lint.
• Finance-firm entity cluster from How the AI revolution has turbocharged M&A: Kkr, Blackstone, Blackrock, Apollo, Calpine, Sandisk, Eaton, General Catalyst. 2 inbound each; not core AI policy. Score 2 each, deferred. (See "Schema clarification needed" — these face the same entity_type ambiguity as the new utility pages.)
• Doctorow-cluster concept stubs: Inverse Cooking Problem, Inverse Trust Problem — may already be covered by OpenClaw / Moltbook (Clawdbot saga); needs scope decision before creation. Score 1 each.

One-line summary

Nine substantive gaps actioned: six new pages live (the Cloudflare-Prince labor-displacement triad — Middle-Manager Displacement, AI Displacement vs. Augmentation, AI Fluency Divide — anchored by Peter Drucker as the framework's 1954 root; and the AI-data-center utility duo Dominion Energy + NextEra Energy capturing the May 18 $67B merger), Institute for Law & AI expanded from 81 to ~400 words, and two foundational primary sources queued for ingest with full authenticity verification — Pope Leo XIV's Magnifica Humanitas encyclical (closing the 2026-05-25 same-day-of-publication deferral) and Topaz et al.'s Lancet correspondence on a 12-fold rise in AI-fabricated biomedical citations.