Research checked 9 September 2026. Root acquisition stopped after the scoped reads below. This packet distinguishes an assessment, an operator's response, current programme reporting and actual case records. No access to private signals was attempted.
[BSR's November 2023 HRIA](http[local research file] PDF19–20, describes a multiparty agreement and Tech Coalition admission, compliance and removal powers. Participants must review signals and establish their own policy violation; they must provide appeals. These are described commitments, not an inspected executed agreement.
BSR warns that decontextualized or erroneous signals can produce cumulative exclusions, requiring separate appeals to each company (PDF25–30). It identifies resource and legal-pressure risks that can undermine review, and dependence on Meta's infrastructure. These are assessed risks, not proven wrongful cases. Its April–August 2023 work predates later financial integration. The report disclaims audit/certification status.
The assessment therefore supplies a testable distinction: distributed decisions can share an evidential input while remedies remain fragmented. The actual quality of investigation and correction propagation remains unresolved.
Access: primary web reader succeeded (63 pages); one ordinary PDF GET returned403. No original PDF retained; do not retry this closed route. Read disclaimer/executive portions PDF2–9, methodology/stakeholder sections PDF14–18, programme/governance PDF19–30, and commitments/recommendations PDF46–50. Remaining pages were not fully read. Reader page excerpts were inspected; this is not full-document substantive review.
The [7 November 2023 response](http[local research file] says nine of nineteen recommendations had been implemented and ten were being pursued or investigated. It describes retention limits, signal-quality review, participation management and planned transparency. Independent audits or trusted researcher access were still possibilities to explore. This is an implementation claim by the commissioning operator, not independent verification of performance.
The main article was read completely. Retained HTML: lantern-hria-response.html,115875 bytes,SHA2564933c787d9a3ff987e4e521e1aaadda5b1a2b3ec512172b9745928b8746b1262. Original acquisition metadata is in root-hria-captures.json.
Root independently read PDF8–11,14,28–29 of the [2025 transparency report](http[local research file] retained by Hubble. Its case study starts with a law-enforcement request to Meta, followed by Meta's investigation and signals, then Cash App closures and Xbox action. This does not equate to government database membership or direct government uploads.
The report describes commissioned BSR reviews of financial participation and AI-related signals, but not their full findings. It states companies may delete only signals they uploaded, and gives aggregate removals without reasons that separate error correction from retention expiry. Own-policy review is required; shared taxonomy can cover conduct prohibited by platform rules without establishing illegality. The listed cross-platform tags include Steam; a tag is not proof of Steam participation or a particular Steam sanction.
Original: lantern-report2025.pdf,836595 bytes,SHA256f6667366d192a4582ab5d6e6ea2115a1772ca3312229a4e48bdc18b68d21bc5d. Hubble's lantern.md carries broader report scope and numerical discrepancies. Do not treat both readings as independent evidentiary origins.
The [2024 report](http[local research file] PDF14, supplies a partial remedy requirement: companies must maintain appeals/recourse, remove signals no longer applicable or relevant, and notify Tech Coalition. Root subsequently read that original page. This establishes a stated removal/notification obligation; the public paragraph does not specify notification to every earlier recipient or reopening its decisions.
The high-value missing artifact is a correction and appeal propagation protocol, with an anonymized worked example: when the originating company retracts a signal, who receives notice, whether downstream investigations are reopened, what happens to already-exported data, and whether affected users learn that several decisions share one source.
The current annual totals cannot answer that question. They neither establish a wrongful exclusion nor demonstrate successful end-to-end remedy. A second useful discriminator is the financial-sector HRIA with its recommendation disposition and actual financial onboarding terms.
Scoped searches: BSR Lantern financial-pilot human-rights review; Tech Coalition Lantern appeals2025; Tech Coalition Lantern financial human-rights. These located a public pilot summary and older/current transparency material, but not the complete later HRIA or an identified public correction case. The [17 July 2025 pilot summary](http[local research file] offers its full report to members/participants through a member resource centre. That access boundary is not authority to enter it; the public summary was read. No outreach or restricted-access acquisition occurred.
Interpretation to carry into the forest: admission and taxonomy authority, infrastructure provision, signal origination, downstream enforcement and remedy are five separate functions. A network may distribute exclusions without one actor issuing each final instruction. To claim that it has done so wrongly requires the actual signal-to-decision-to-correction evidence.