AI Policy Wiki
Dashboard
⚲
AI Policy Wiki › Briefings

Gap Scan — 2026-06-12

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

Contents

Scanned

Recent window: 4 dev-log files (2026-06-10-1513 through 2026-06-11-2215; the 06-11-2215 evening file is not yet folded by the nightly cycle and got recent-thread treatment only), ~43 content pages edited in last 48h. Rotation slice: 8 — litigation/ (18 pages) + standards/ (6) + government/ (15), scanned for anchor-thinness, in-degree, and stale last_status_check.

Gaps actioned (9 of ~20 found)

New pages created (live)

  • LG München I — Google AI Overviews liability (26 O 869/26) — carried from the 06-11 deferred backlog (score 5, "re-evaluate next run"); the nightly cycle folded the event into Google DeepMind but the case itself — the first German decision prohibiting specific false AI Overview statements, with Google held a direct disturber whose AI output is its own content — had no litigation/ page despite sitting squarely in this run's rotation slice. Built from heise (which carries the JUVE cost-allocation detail and the LG Frankfurt 2-06 O 271/25 lineage) + the-decoder + the Marcus Section-230 commentary. Plaintiff publishers are not publicly named; noted honestly in frontmatter. confidence: medium, 3 sources. The primary judgment text is not yet on a public canonical host — flagged below for a future queue.

Pages expanded (live)

  • DOGE — Department of Government Efficiency (AI Deployer) — slice-8 thin anchor (270w × in-degree 6, the shortest qualifying page in government/). Story had moved substantially past the page's mid-2025 state: added the January 2025 establishing EO (whitehouse.gov), the May 21, 2026 Reuters report that Grok saw minimal government adoption and that DOGE — now described as defunct — had pushed DHS to use unapproved Grok, the February 2026 nutrition-website-to-Grok item, Warren's March 2026 Pentagon/xAI classified-networks letter, and a Status and wind-down section (GSA departure + IG efficiency review, Sept 2025). [5 new (Source: URL) cites] sources_count 3 → 8.
  • NIST AI Agent Standards Initiative (2026) — slice-8 thin anchor (839w × in-degree 19, sources_count: 1). Added the canonical nist.gov initiative-page citation (created Feb 17, updated Apr 20, 2026) including the stated "cementing U.S. dominance at the technological frontier" goal, the RFI issuance date (January 2026), the listening-session registration deadline (March 20), and corrected the NCCoE concept-paper row: published February 5, 2026 (CSA research note), with April 2, 2026 the comment deadline, not the release date. [2 new sources] sources_count 1 → 3.
  • Raine v. OpenAI, Inc. — highest-in-degree litigation page (72) with a 6-week-stale last_status_check (2026-04-28), live-thread-adjacent (the June 11 Carrier filing in the unprocessed evening dev-log references the coordinated proceeding). Added a Related suits and coordinated proceeding section: November 2025 seven-suit wave (total 8), the 18-suit coordinated California proceeding as of June 11, 2026 (Reuters), and the Carrier filing seeking mandatory self-harm-conversation termination. last_status_check → 2026-06-12; sources_count 3 → 5.

Queued — foundational sources

  • Narayanan & Kapoor, "Why AI hasn't replaced software engineers, and won't" (June 10, 2026) — carried from 06-11 (score 5, over cap). Advances the decide-execute-deliver-sandwich framework, the AI-washing case studies (Block/Snap/Intuit), and original NY WARN-act reporting. Verified: normaltech.ai (the authors' own venue; full text scraped, distinctive datapoints match the dev-log fold). Saved: Raw Sources/Why AI Hasn't Replaced Software Engineers (Narayanan & Kapoor, 10 June 2026).md (full verbatim text). Queued: INGEST-narayanan-kapoor-software-engineers-2026.md.
  • Anthropic "Advanced AI Framework" (June 2026) — lint backlog (06-11 suggested sources) + carried deferral ("queue once canonical URLs confirmed"). Verified: anthropic.com landing page (scraped, HTTP 200) + full-text PDF on www-cdn.anthropic.com; thresholds and four risk areas match the ingested Amodei essay. Large-PDF protocol: fetch fresh at ingest. Queued: INGEST-anthropic-advanced-ai-framework-2026.md.
  • Anthropic "Economic Policy Framework" (June 2026) — same gap pair; would confirm the $200M/$150M split and universal-capital-accounts mechanics now sourced to trade press. Verified: anthropic.com landing page names and links it (/policy-on-the-ai-exponential/epf); subpage deliberately not pulled to respect the daily web-pull cap — confirm full text at ingest. Queued: INGEST-anthropic-economic-policy-framework-2026.md.

Authenticity-verification failures

  • None. Both pulled texts verified on canonical hosts (normaltech.ai; anthropic.com); the two Anthropic framework documents were verified at landing-page level on the issuer's own domain with full-text retrieval deferred to ingest.

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

  • Munich primary judgment text (26 O 869/26) — lint-suggested foundational ingest; not yet found on a public canonical host (German PI decisions in press-law matters often go unpublished). Re-check gesetze-bayern.de/openjur next runs. Score 4.
  • EU Code of Practice on Transparency of AI-Generated Content (full text) — lint-suggested; ec.europa.eu host known. Score 4 — strong queue candidate next run.
  • June 11 evening dev-log items (Carrier suit page-worthiness, JAWBONE Act, Kelly S. 4742, Helix Digital Infrastructure/KKR, Robotic and Autonomous Systems Combatant Command NDAA provision, OpenAI–Ona acquisition) — nightly developments-log fold gets first pass; the Carrier suit is now contextualized on Raine v. OpenAI, Inc. either way. Re-evaluate gaps after the fold.
  • Slice-8 thin anchors not reached: Five Eyes Joint Guidance on Secure Deployment of AI Agents (May 2026) (419w × 6, src 1), Character.AI Litigation (umbrella) (742w × 6, sources_count: 0 against confidence: high — frontmatter inconsistency, route to lint), Garcia v. Character Technologies, Inc. (deg 66, status check 2026-04-28) and NYT v. Microsoft, OpenAI et al. (deg 27, status check 2026-04-25) — stale last_status_check, candidates for the next litigation status pass. Score 2–3 each.
  • claude-code (in-degree 8) — perennial since 06-04; still blocked on the product-page placement decision. Needs a schema call from the user. Score 3.
  • entities/mike-rounds (2+), concepts/sovereign-ai / concepts/safetywashing / concepts/regulatory-managerialism (2 each) — small dangling concept/entity targets. Score 1–2.
  • Lint territory (route to lint, not gap-scan): sources/anthropic-pwc-expanded-partnership (6) and sources/ft-mundy-americans-dread-ai-2026 (3) citation-format conversions; case_status vs status field drift on litigation frontmatter; the \| table-alias scanner escape (flagged 06-11).

Failures

  • Dashboard rebuild failed (sixth consecutive run): bin/build-dashboard.py aborts at rmtree on the iCloud sync-conflict * 2 directories inside Wiki/_meta/site/ (PermissionError: 'companies 2'). Host-side fix required — already documented in queue/gap-scan/needs-review/2026-06-11-dashboard-rebuild-failed.md. Content graph unaffected.

One-line summary

One new litigation page (the Munich AI Overviews direct-liability ruling), three expansions (DOGE wind-down, NIST agent-standards sourcing, Raine coordinated-proceeding status), seven alias instances fixed, and three foundational sources verified and queued (Narayanan/Kapoor essay + both Anthropic policy frameworks) — the queue now holds the full primary-text set behind the June 10 Anthropic policy push, awaiting your review.

Categories: Briefings | briefing | gap-scan