BCHRT decision-access audit

01

Established from current sources

The public can read and search BCHRT decisions.

The Tribunal directs users to CanLII and lists decisions by year from 1997 onward. It warns that the signed original in the Tribunal file is authoritative. Its copyright page permits reproduction subject to accuracy, currency, source and non-affiliation conditions; commercial multiple reproduction requires permission.

The Tribunal expects parties to work with legal authorities and asks for neutral citations and free-access versions where possible. Its self-represented guide confirms that people may represent themselves while explaining that lawyers analyze evidence and case law.

SourcesBC Human Rights Tribunal · BC Human Rights Tribunal · BC Human Rights Tribunal · BC Human Rights Tribunal

02

Adverse finding

Search+ is public. That does not settle the access question.

CanLII launched Search+ on February 25, 2026 for anyone with a free myCanLII account. The project therefore will not claim that AI-assisted CanLII search is available only to lawyers.

The deeper issue is not whether one interface currently works. It is whether the public has a stable, authorized route to retrieve, preserve, reuse and computationally analyze public tribunal decisions without depending on one intermediary’s interface, terms, model and continued willingness to provide the service.

SourcesCanLII Blog · CanLII · CanLII

03

What Search+ currently provides

A helpful research assistant, with a defined ceiling.

The current public limits are ten AI-generated search queries and four Analyze runs per day. Each Analyze run reads only the first 25 results, for a maximum of 100 analyzed documents per day. The limits may change.

The product page says users can analyze thousands of legal documents in seconds. The operating description is narrower: Search+ searches CanLII and its Analyze function reads the first 25 results. This audit does not interpret the marketing line as full-text LLM analysis of thousands of documents in one run.

SourcesCanLII · CanLII · CanLII

04

The structural issue

Even unlimited Search+ would not create public data access.

A larger quota would make today’s interface more convenient. It would not give the public an authorized copy of the corpus, the right to preserve versions, the ability to choose or evaluate a different model, or a stable route that survives a product change or closed entrance.

Public access cannot depend on the continued availability of one website, one search interface or one approved AI model. CanLII can be one valuable service built on public decisions. It should not have to be the only practical gate through which the public can computationally use them.

SourcesBC Human Rights Tribunal · CanLII · CanLII

05

Privacy and reproducibility

A person cannot safely treat Search+ as a private case workspace.

CanLII’s terms prohibit sensitive, confidential, privileged or personal information in Search+ queries. Its privacy policy says service providers may process query content under contract, but confidentiality is not guaranteed. An unredacted case narrative, medical record or privileged communication therefore does not belong in the tool.

The official documentation also conflicts on history. The current Search+ FAQ says new history is retained in the myCanLII account across devices and after local data is cleared. The March 2026 Privacy Policy says Search+ interactions are stored only in the browser and are not retained in CanLII’s systems. Until that conflict is resolved, reproducible research must keep its own non-sensitive query and result record.

SourcesCanLII · CanLII · CanLII

06

The unresolved infrastructure gap

Public viewing does not authorize systematic retrieval.

CanLII’s terms prohibit systematic downloading, including through crawlers, plugins, connectors, bots or organized manual collection. Automated or large-scale retrieval is directed to original sources or authorized channels.

CanLII ran a commercial API pilot in 2022 that contemplated model training in a particular program. It does not establish that a current BCHRT API or licence is available to this project. Current public Search+ materials do not document a batch-analysis API, a complete result export or a corpus download.

This is an active governance boundary, not a theoretical footnote. CanLII publicly announced legal action in 2024 over alleged unauthorized data collection; the proceeding was resolved in 2026 on confidential terms. The settlement announcement does not create a public retrieval right.

SourcesCanLII · CanLII Blog · CanLII Blog · CanLII Blog

07

Questions still open

The first development milestone is authorization, not scraping.

• Which decisions, interim decisions, corrections and reconsiderations exist in each source?
• Which file is the latest or authoritative public version, and how are replacements recorded?
• Can citation, date, member, grounds, area, outcome and referenced authorities be obtained reliably?
• Does BCHRT offer an export, feed, API or written permission for public-interest indexing?
• What rights apply to full-text storage, excerpts, versioning and model processing?
• How will the tool disclose what it did not index and identify relevant decisions a search missed?

08

Audit deliverable

The output will be a public source ledger, not a private workaround.

The ledger will identify every source, retrieval method, date, terms version, document count, exclusion, version marker, correction process and written authorization. Failed and blocked routes stay in the record.