AI Policy Wiki
Dashboard
⚲
AI Policy Wiki › Briefings

Gap Scan — 2026-07-13

Daily gap hunt — what the wiki is missing and how each gap was triaged.

Contents

Scanned

Recent window: 4 dev-log files (07-11 morning/evening, 07-12 morning/evening), ~43 pages edited in last 48h. Rotation slice: 11 — sources/ A–H (243 pages; first pass over this slice since the daily reports began). Note: the 07-12 evening dev-log (22:05) remains unprocessed for tonight's developments-log cycle; only its gap-relevant items were actioned here (07-10/07-11/07-12 precedent) — its non-gap folds (Heidecke departure/safety reorg, Fable 5 access extension, coding-model price war, Boko Haram chatbot use, Pangram LinkedIn study, Indeed Hiring Lab postings data, Cursor "Sand", Ray Sun 50-state tracker) were left for the nightly cycle.

Gaps actioned (7 of ~17 found)

New pages created (live)

  • Stop AI — top live-thread gap (score ~6: +3 live 07-12 evening dev-log, type-3 missing entity page, recurrence condition met). Deferred on 07-11 and 07-12 as "page if references recur" (Markey precedent) — the 07-12 evening WSJ profile is the second touch in 48h, and the group was already named in prose on Public Opinion on AI's Organized-opposition section with no page behind it. Built from The Register (2025 AGI-ban goal, primary coverage of the group's founding position), Fortune's April 2026 feature (founding by Kirchner and Reichstadter, PauseAI split, Moreno-Gama forum episode, five-co-leader structure), SFGate (Altman subpoena), SF Standard (November 2025 OpenAI lockdown), KQED (Reichstadter arrest), SF Gazetteer (Kaufmyn trial), Daily Californian (strategy rethink), WSJ 07-12, and Politico. Radicalization commentary presented both ways (Weiss-Blatt vs. Lubrano), attributed. Backlinked from Public Opinion on AI (existing mention wikilinked), PauseAI (new sourced Relationships line), and Index. confidence: medium.

Pages expanded (live)

  • Bartz v. Anthropic — deferred carry (07-12) turned live: stale since 06-06 with last_status_check 05-11 while the May 14 fairness hearing record sat only on AI Copyright Litigation — Analysis. Added a Settlement administration and final approval section (opt-out/re-inclusion/claims deadlines; the 75-minute May 14 hearing before Judge Martínez-Olguín — 92.77% claims rate, seven objectors, fees/cost-reserve questioning, May 21 supplemental brief on late opt-outs; "minuscule" opt-out rates per Publishing Perspectives; approval still pending per Clark Hill June), the ~500,000-title exposure and preliminary-approval record, the 28-author opt-out jury-trial suit, an updated infobox and two new Open questions. Replaced the stub self-reference note with a provenance note. sources_count 3 → 9; last_status_check → 2026-07-13. [7 new (Source: URL) cites]
  • Sora and Veo (Video Generation Models) — twice-carried thin-frontmatter anchor (high on sources_count: 2 at in-degree 21, stale 06-06, missing parameters/safety_case) — and corroboration research found the page's core framing outdated: OpenAI announced Sora's shutdown March 24, 2026 (WSJ-broken, confirmed by OpenAI; "side quests" refocus; Disney $1B/200-character deal left unclear), Sora no longer generally available by late June; Google countered with Veo 3.1 Lite (March 31, half-cost API) and Veo-based Genie world models; Alibaba's HappyHorse 1.1 reached No. 2 in global video rankings by June as Sora and Seedance fell. Lead, infobox, and Market-entry sections updated; frontmatter repaired. sources_count 2 → 7. [6 new (Source: URL) cites]
  • The Bletchley Declaration (AI Safety Summit, 1–2 November 2023) — slice thin anchor #1: high on sources_count: 1 at high in-degree (the stochastic-parrots pattern). Corroborated against the GOV.UK primary text: full signatory list (28 countries + EU), New Zealand's October 23, 2024 accession (new), the primary text's frontier-AI definition and cybersecurity/biotechnology risk domains, the "meeting again in 2024" closing line, and the publishing provenance (PMO/FCDO/DSIT, Open Government Licence). source_url added. sources_count 1 → 2. [3 new (Source: URL) cites, all gov.uk]
  • EU General-Purpose AI Code of Practice (Final Version, 2025) — slice thin anchor #2: high on sources_count: 1 at in-degree ~95. Corroborated against the Commission's own digital-strategy.ec.europa.eu page (updated April 23, 2026): the 23-name signatory list presented alongside the page's prior "more than 27" claim (EC caveats continuous updating; both retained, attributed), the EC's exact xAI alternative-adequate-means language, and the Signatory Taskforce / Vademecum structure. source_url added. sources_count 1 → 2. [3 new (Source: URL) cites]
  • People-First Chatbot Act — carried queue candidate resolved as an expansion: the bill number was unidentified on the wiki ("committee referral… had not been reported"). Verified on Congress.gov: H.R. 9619, official title, July 9 introduction, referral to House Energy and Commerce — but the bill text is not yet posted, so the primary-text ingest cannot be queued yet (see below). Added bill number, referral, official title, the Foushee-office 30+-endorsers framing (primary host), and the CFA endorsement. sources_count 2 → 5. [3 new (Source: URL) cites]

Queued — foundational sources

  • None this run. The one queue candidate that matured — the People-First Chatbot Act primary text — failed the full-text gate: Congress.gov carries the bill record (H.R. 9619) but "text becomes available" is still pending, so queueing an ingest now would point at a document that does not yet exist. The verified congress.gov URL is on the legislation page; queue the text ingest when it posts (typically within days of introduction).

Authenticity-verification failures

  • None. All pulled supporting sources verified on their expected hosts (gov.uk, digital-strategy.ec.europa.eu, congress.gov, foushee.house.gov, plus mainstream outlets for Stop AI and Sora). The H.R. 9619 text-pending situation is recorded above as a deferral, not a failure.

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

  • US AI Regulatory Approaches Compared (in-deg 36, medium, stale 06-06, missing sources_count) and Labor Disruption Timelines: Who Predicts What and Why (in-deg 21, same pattern) — carried from 07-12; each needs a substantive multi-instrument update, deferred for a run with capacity. Score ~4 each.
  • Nathan Lambert, "6 months to live for open models" (Interconnects, 07-12) — essay advancing an original argument on a live thread (reported White House open-weight EO discussions); queue candidate once tonight's cycle folds the thread and a wiki page names it. Score ~3.
  • Benedict Evans, "Ways to think about token pricing" (07-09; carried by the 07-12 evening dev-log) — same pattern as the already-queued Evans essay; queue candidate after the nightly fold. Score ~3.
  • Pangram Labs social-media AI-content study (1,002,627 posts; 07-12 evening) — original empirical claim; nightly fold first, borderline foundational. Score ~3.
  • H.R. 9619 bill text ingest — blocked on Congress.gov text posting; re-check next run. Score ~3.
  • Slice-11 residue (two-per-run pacing): sources/hiroshima-code-of-conduct (high/1, in-deg ~55), sources/alignment-faking-paper (high/1, ~53), sources/china-generative-ai-interim-measures (high/1, ~60), sources/eo-14110 source page (high/1; the legislation twin was fixed 07-09), sources/ai-2027 (medium/1 at in-deg 64 — source-robustness-check territory rather than gap-scan). Score ~3 each.
  • Possible standalone models/gemini-3-5 page once 3.5 Pro ships (reported July 17). Score ~3.
  • Helix (Figure AI Vision-Language-Action model) (257w, medium/1, in-deg 6) — carried; confirm next models slice. Score ~3.
  • Markey "AI Accountability Agenda" primary document — carried. Score ~2.
  • Standing carries: claude-code/claude-cowork placement (curator-blocked), inverse-cooking/inverse-trust coined terms (needs-review), duplicate entities/cdao/government/cdao, concepts/foundation-models disambiguation, person/org pages at in-degree 2, ChinaTalk Eastern-Data-Western-Compute concept page (folded 07-11; page only if the thread recurs). Score ~2–3.

One-line summary

Stop AI finally has an entity page after its third live touch in 72 hours, the Bartz case page caught up to the May fairness-hearing record it was missing, corroboration research on the twice-carried sora-veo anchor surfaced that Sora was discontinued in March 2026 and rewrote the page's framing, the slice's two weakest high-traffic primary-text source pages (Bletchley, GPAI Code of Practice) now carry primary-host corroboration, and the People-First Chatbot Act was pinned to H.R. 9619 — nothing queued for review this run; the only pending user action remains the existing ingest queue.

Categories: Briefings | briefing | gap-scan