AI Policy Wiki
Dashboard
⚲
AI Policy Wiki › Briefings

Gap Scan — 2026-07-01

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

Contents

Scanned

Recent window: 2 dev-log files (2026-06-30-1939, 2026-07-01-0510) and ~23 content pages edited in the last 48h (notably models/gpt-56, models/glm-5, models/claude-fable-5, models/claude-mythos-5, companies/openai, companies/anthropic, companies/microsoft, concepts/export-controls-ai, concepts/ai-bubble-debate, legislation/ai-data-center-moratorium-act, overview.md). Rotation slice: 13 — sources/ Q–Z (≈95 pages) + Raw Sources/ orphan check. Methods: wiki-wide broken-link analysis (existing-slug subtraction + inbound ranking with table-escape filtering), alias resolution on every candidate, a recent-thread scan of the two dev-logs, and an anchor-thinness cross of the slice against in-degree. Deduplicated against the open queue (57 INGEST- tasks + proposed-sources/ staging), the gap-scan reports from the last 7 days (06-24 → 06-30), and existing Raw Sources/ files.

Gaps actioned (7 of ~11 surviving candidates)

New pages created (live)

  • Claude Sonnet 5 — top live-thread gap and a wiki core area (frontier models / frontier-lab governance). Anthropic released Claude Sonnet 5 on June 30, 2026 — the largest single item in both the 06-30 and 07-01 dev-logs — with no page. Built from Anthropic's primary announcement (scraped live), the Sonnet 5 System Card references within it, and Axios. Frontmatter set for the models schema (developer, release_date, open_weights, safety_case). confidence: medium. [Depends on queued source: Claude Sonnet 5 System Card — cited inline with a Provenance note.]
  • Kids Internet and Digital Safety Act (KIDS Act, H.R. 7757) — live-thread gap (House passed H.R. 7757, 267–117, June 29, 2026); no youth-online-safety legislation page existed (KOSA/COPPA/KIDS all absent). AI relevance is the SAFE Bots Act title (chatbot safeguards for minors). Built from Pillsbury's provision analysis (scraped), K-12 Dive, 5 Calls, and the H.R. 7757 bill text. confidence: medium.
  • ICTS Supply Chain Security Act — live-thread gap in a core area (export controls / national security / compute). Senate bill introduced June 30, 2026 by Sens. Scott and Hagerty to expand Commerce's foreign-adversary ICTS authority while preserving open-source AI access. Built from CoinDesk (scraped) and the Senate Banking Committee bill text. confidence: medium.
  • Tom Brown — structural gap: broken [[entities/tom-brown]] at in-degree 6 (referenced substantively in companies/anthropic, concepts/export-controls-ai, models/claude-mythos-5, models/claude-fable-5) with no page. Anthropic co-founder and chief compute officer; GPT-3 lead author at OpenAI; lead in the June 2026 White House export-control negotiations. Built from Forbes, Bloomberg, NeuronFeed, LinkedIn, plus the Verge/Wired reporting already cited on existing pages. confidence: medium.

Pages expanded (live)

  • EU AI Act (Regulation 2024/1689) — the "Reform and amendment activity (2026)" section stopped at the 20 May 2026 Digital Omnibus political agreement. Folded the 29 June 2026 Council final approval ("Omnibus VII"): confirmed high-risk deferral to 2 Dec 2027 / 2 Aug 2028, and the new prohibition on AI generating CSAM or non-consensual intimate content applying 2 Dec 2026. Updated the application-timeline note and last_updated. [2 new (Source: URL) cites — Baker Botts, Sidley.]
  • AI Labor Disruption — folded the Ramp/Revelio Labs study (June 30, 2026) into the "Hiring composition and new roles" section alongside the SignalFire datapoint: heaviest AI spenders grew headcount ~10% and entry-level roles ~12% over two years, with Ramp's lead economist cautioning correlation-not-causation. sources_count 41 → 43. [2 new (Source: URL) cites — Big Technology, CoinDesk.]
  • None this run. The broken-link scan surfaced no clean alias correction; the top broken targets were either operational (lint-report, log, briefings/…), table-escape artifacts (companies/anthropic\, claude-mythos-5\), or previously-deferred items (see Deferred backlog).

Queued — foundational sources

  • Claude Sonnet 5 System Card (Anthropic, 2026-06-30) — foundational safety artifact for the new Claude Sonnet 5 page. Verified: anthropic.com (landing) + www-cdn.anthropic.com (PDF), both linked by name and by section reference (§3.2.4, §6.4) from Anthropic's own announcement. Large PDF → fetched fresh at ingest (no pre-saved Raw Sources/ copy, per the QUEUE rule). Queued: INGEST-claude-sonnet-5-system-card-2026-06-30.md; verification record in proposed-sources/claude-sonnet-5-system-card-2026.md.

Authenticity-verification failures

  • None. One foundational source verified and queued (Sonnet 5 System Card). All other new/expanded pages were built from supporting web sources folded as (Source: URL); no other primary documents were downloaded.

Deduplicated (already pending / already covered — not re-actioned)

Deferred backlog (over the daily cap / out of lane — re-surfaces next run)

  • OpenAI "Jalapeño" custom inference chip (Broadcom-co-developed, unveiled June 24) and the ~50% inference-cost optimization — recent-thread items; best folded into OpenAI / a compute page by the nightly cycle rather than a new page. Score ~4.
  • South Korea $576B AI-chip drive (Samsung / SK Hynix pledges, June 29) — folds into SK Hynix — HBM Leader / a compute page as supporting updates, not a new page. Score ~4.
  • Google "Personal Intelligence" / Nano Banana free-tier expansion (June 29) — update-class for a Gemini/Google page; supporting news-cycle item. Score ~3.
  • Thin high-in-degree sources/ Q–Z anchors → source-robustness-check, not gap-identifier: e.g. The Bitter Lesson, SWE-bench: Can Language Models Resolve Real-World GitHub Issues?, Superintelligence: Paths, Dangers, Strategies — Bostrom (2014), Scaling Laws for Neural Language Models are heavily-linked source summaries whose faithful deepening should come from their Raw Sources/ files. Out of gap-identifier lane. Score ~4 each (flagged for the robustness skill / lint).
  • [[claude-code]] (in-degree 7, broken) — perennial; placement genuinely ambiguous (product vs. concept); needs a deliberate products/-schema call from the user. Score ~4.
  • sources/anthropic-pwc-expanded-partnership (in-degree 6, broken) — reads as a supporting partnership announcement; should fold into Anthropic (a lint/fold task, no sources/ page warranted). Score ~3.
  • Person pages at in-degree 2 — entities/masahiro-mori, entities/anil-seth, entities/eric-horvitz, entities/orin-kerr, entities/alondra-nelson, entities/clayton-christensen, entities/palmer-luckey. Low single-scan value; confirm substance before creating. Score ~3 each.

Notes for lint

  • legislation/eu-ai-act has no sources_count frontmatter — a frontmatter repair (the page carries 30+ citations). Not added this run to avoid an inaccurate count; flagged for lint.
  • Pre-existing broken links in concepts/ai-labor-disruption: [[companies/google]] and [[entities/cac]] were already present (not introduced by this run's edit) — candidates for a future alias/creation pass (companies/google likely a genuine missing page or alias to an existing Google/Alphabet page; entities/cac is the Cyberspace Administration of China).
  • Stray temp file: check_links.py (~1 KB) was written to the vault root during this run's broken-link verification and could not be removed by the sandbox ("Operation not permitted" on the iCloud-synced path). It sits outside Wiki/, so it does not affect content indexes or the dashboard build. Safe for the user to delete manually.

One-line summary

Four live pages created — Claude Sonnet 5 (Anthropic's June 30 mid-tier agentic model), Kids Internet and Digital Safety Act (KIDS Act, H.R. 7757) (H.R. 7757, House-passed June 29, with the SAFE Bots AI-chatbot title), ICTS Supply Chain Security Act (Scott/Hagerty, introduced June 30), and Tom Brown (Anthropic chief compute officer, in-degree 6) — plus two thematic folds (EU AI Act (Regulation 2024/1689) June 29 Council final approval; AI Labor Disruption Ramp/Revelio study) and one foundational source (Claude Sonnet 5 System Card) verified and queued for review.

Categories: Briefings | briefing | gap-scan