Gap Scan — 2026-06-15

Scanned

Recent window: 2 dev-log files in the last 48h (2026-06-14-2118, 2026-06-14-2205 — the Anthropic Mythos/Fable export-control episode plus a children's-AI litigation/legislation cluster); 9 wiki pages edited in the last 48h (the software-engineering-labor thread carried over from 06-14: companies/dropbox, companies/notion, concepts/ai-coding-agents, concepts/ai-labor-disruption, concepts/effective-altruism, concepts/regulatory-managerialism, entities/arvind-narayanan, entities/sayash-kapoor, sources/why-ai-hasnt-replaced-software-engineers). Rotation slice: 11 — sources/ A–H.

The dominant live thread (the June 12 Lutnick export-control directive on Mythos 5 / Fable 5) is breaking news whose primary artifacts — NSPM-11, Amodei's "Policy on the AI Exponential," the Lutnick letter — are already staged in queue/gap-scan/proposed-sources/ or ingested (sources/amodei-policy-on-the-ai-exponential); folding the episode itself is developments-log's job, so no page was created for it here. The children's-AI litigation named in the dev-log (Florida v. OpenAI, the Character.AI cases) was found already covered in litigation/. The freshly created concepts/effective-altruism and concepts/longtermism pages left [[entities/will-macaskill]] dangling, which this run resolved.

Gaps actioned (6 of ~11 found)

New pages created (live)

• Safetywashing — dangling [[concepts/safetywashing]] (in-degree 3: Sociotechnical AI Risk Governance, Regulatory Managerialism which marks it *(stub)*, and sources/knight-columbia-mulligan-marda-wang-risk-governance-2026-03). Core safety/measurement concept. Anchored to Ren et al., "Safetywashing: Do AI Safety Benchmarks Actually Measure Safety Progress?" (arXiv:2407.21792, NeurIPS 2024). confidence: medium.
• William MacAskill — dangling [[entities/will-macaskill]] newly relied on by the 06-14 concepts/effective-altruism and concepts/longtermism pages (both mark it *(stub)*). Central EA/longtermism figure. Built from MacAskill's own site and Substack plus reference material. confidence: medium.
• New York State Department of Financial Services (NYDFS) — dangling [[entities/ny-dfs]] (in-degree 2 in CISA — Cybersecurity and Infrastructure Security Agency (AI Deployer)); a state financial regulator that issued a May 2026 frontier-AI cybersecurity advisory. US-regulation core area. Anchored to the primary NYDFS industry letter (dfs.ny.gov) plus Sidley/Greenberg Traurig analyses. confidence: medium.
• Handoffs (sociotechnical concept) — dangling [[concepts/handoffs]] (in-degree 2: Sociotechnical AI Risk Governance, Deirdre K. Mulligan); the Mulligan–Nissenbaum "handoff" lens that anchors the sociotechnical-governance cluster. Anchored to "The Concept of Handoff as a Model for Ethical Analysis and Design" (SSRN 3784839). confidence: medium.
• Digital divide — dangling [[concepts/digital-divide]] named as the "general concept" parent of AI Divides (Literacy / Occupational / Ethico-Philosophical) and referenced by sources/project-deal and companies/anthropic (the "second digital divide" framing). Built from UNESCO, ILO, a peer-reviewed AI-divide study, and Brookings. confidence: medium.

Pages expanded (live)

• None this run. The rotation slice (sources/ A–H) and the live-thread anchor pages were already adequately built; the run's budget went to the five missing-page gaps and the alias fix.

Link aliases fixed

• [[machines-of-loving-grace-amodei]] → [[machines-of-loving-grace]] (2 instances in sources/andersen-dario-amodei-oppenheimer-moment.md). The Amodei essay already has a sources/machines-of-loving-grace page; the -amodei suffix was a link alias, not a missing page.

Queued — foundational sources

• None new this run. The foundational works surfaced are already pending: how-will-openai-compete-evans and a-vision-of-democratic-ai both already have Raw Sources/ files (deduped); trustworthy-agents-in-practice is already queued (INGEST-trustworthy-agents-in-practice-anthropic.md); the live-thread primaries (NSPM-11, Amodei "Policy on the AI Exponential") are staged or ingested. Three of the five new pages above carry a provenance note marking a foundational ingest that would deepen them (Ren et al. safetywashing paper; MacAskill's own writing; the Mulligan–Nissenbaum handoff chapter) — candidates for a future queue cycle, not pulled today to stay within cap.

Authenticity-verification failures

• None. No raw sources were pulled or saved this run; all new pages were built from supporting web sources cited inline as (Source: <URL>), per the AUTO new-page lane. The NYDFS letter and the safetywashing arXiv record were verified against their canonical primary hosts (dfs.ny.gov; arxiv.org, DOI 10.48550/arXiv.2407.21792) for citation accuracy but were not saved as raw sources.

Data discrepancy flagged for lint

• NYDFS advisory date. CISA — Cybersecurity and Infrastructure Security Agency (AI Deployer) cites the frontier-AI advisory as "May 26, 2026." The primary NYDFS industry letter is dated May 21, 2026 (https://www.dfs.ny.gov/industry-guidance/industry-letters/20260521-heightened-cybersecurity-risks-assoc-with-frontier-ai-models). The new entities/ny-dfs page uses the primary date; government/cisa should be corrected by lint or the curator.

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

• claude-code — in-degree 8. Perennial since 2026-06-04; blocked on a product-page placement schema decision (no products/ folder). Needs a schema call from the user — not auto-created. Score 3.
• sources/anthropic-pwc-expanded-partnership (in-degree 6) — supporting-class partnership announcement mis-linked as a sources/ page; should be folded as inline (Source:) citations. Hand to lint. Score 2.
• concepts/frontier-models — a general "frontier models" definitional page distinct from concepts/open-weight-frontier-models / concepts/frontier-ai-governance. Score 2.
• concepts/a-vision-of-democratic-ai — linked from concepts/ai-and-democracy as a concept, but a Raw Sources/A Vision of Democratic AI.md file exists; this is a source, not a concept. Resolve once ingested (link should point to sources/). Score 2.
• concepts/sovereign-ai alias question (→ a governance-sense page distinct from the existing concepts/sovereign-ai-product?). Score 1–2.
• In-degree-2 entity stubs referenced in essays: entities/pam-bondi, entities/soufan-center, entities/recalc-academy, entities/masahiro-mori, entities/anil-seth, entities/lee-anne-fennell, entities/richard-dawkins, entities/adam-smith. Score 1.
• Concept stubs: concepts/ai-and-relationships, concepts/inverse-cooking-problem, concepts/inverse-trust-problem. Score 1.

Dashboard rebuild — blocked (environment, 6th consecutive run)

python3 bin/build-dashboard.py again failed at shutil.rmtree(SITE) with PermissionError: [Errno 1] Operation not permitted: 'companies 2'. Eight empty iCloud sync-conflict directories remain under Wiki/_meta/site/ — companies 2, concepts 2, entities 2, industries 2, models 2, primers 2, sources 2, standards 2 — and cannot be removed, moved, or rmdir'd from the sandbox (macOS/iCloud permission restriction). The 5 new pages and the alias fix are live and correct; only the generated HTML site is stale. Action for the user: delete the eight stray "<folder> 2" conflict directories under Wiki/_meta/site/ from Finder, then re-run python3 bin/build-dashboard.py. This has recurred on 06-06, 06-08, 06-10, 06-11, 06-14, and today — worth resolving once at the source. A needs-review note is filed.

One-line summary

Created 5 pages live (Safetywashing, Will MacAskill, NYDFS, Handoffs, Digital divide) and fixed 1 link alias (2 instances); no sources pulled or queued. Flagged a date discrepancy on the NYDFS advisory in government/cisa for lint, and the recurring dashboard-rebuild blocker (iCloud conflict dirs) needs the user to clear it from Finder.