Lead story
ESI search terms do not replace Rule 34 duties
A federal court rejected a party's argument that an ESI protocol limited its email collection to seven search terms and ordered a broader search for responsive material.
- What happened
- A federal court rejected a party's argument that an ESI protocol limited its email collection to seven search terms and ordered a broader search for responsive material.
- Why it matters
- Counsel cannot delegate the duty to identify responsive documents to an opponent's keyword list. Protocols should state that terms, custodians, and workflows implement discovery requests without narrowing independent collection duties, and teams should document follow-up when results conflict with known facts or expected records.
- Watch
- The October 29 article analyzes an October 20, 2026 discovery order reported at 2026 WL 3135943. The cited source is practitioner analysis; the full order was not available through an open canonical court page located during this run.
Sourcesedrm.net
Briefing item
AI chats enter preservation and privilege analysis
New legal guidance treats relevant prompts, outputs, account settings, and chat histories as ordinary ESI while distinguishing possible work-product protection from waiver and consumer-tool disclosure risks.
- What happened
- New legal guidance treats relevant prompts, outputs, account settings, and chat histories as ordinary ESI while distinguishing possible work-product protection from waiver and consumer-tool disclosure risks.
- Why it matters
- Legal teams should inventory approved and unapproved AI services, map retention and export controls, update hold language, and decide when counsel-directed use requires segregated workspaces. Before citing or relying on an AI output, counsel should assess whether doing so may put the related prompts and testing history at issue.
- Watch
- The October 29 article synthesizes existing rules and decisions rather than reporting a new holding. Protection remains fact-specific, and the analysis distinguishes open consumer tools from confidential enterprise or internal systems.
Sourcesnatlawreview.com
Briefing item
Deepfake evidence puts provenance ahead of appearance
Canadian litigators say existing authentication rules are being tested by both synthetic evidence and claims that authentic recordings are fake, increasing the importance of metadata and chain of custody.
- What happened
- Canadian litigators say existing authentication rules are being tested by both synthetic evidence and claims that authentic recordings are fake, increasing the importance of metadata and chain of custody.
- Why it matters
- Collection plans for consequential media should preserve native files, device and platform metadata, transfer records, hashes, and available originals. Review protocols should define the factual showing needed to trigger enhanced authentication work and retain enough provenance to rebut both manipulation and unfounded deepfake allegations.
- Watch
- This is practitioner reporting, not a new Canadian ruling or enacted deepfake-specific authentication standard. The article notes that reported cases may understate undetected manipulation and that courts must calibrate scrutiny to avoid unnecessary forensic expense.
Sourcescanadianlawyermag.com
Briefing item
California brings health chatbots under medical-information rules
California's AB 1979 treats qualifying consumer health chatbots as health-care providers under the CMIA and adds limits on AI-supported clinical functions beginning January 1, 2027.
- What happened
- California's AB 1979 treats qualifying consumer health chatbots as health-care providers under the CMIA and adds limits on AI-supported clinical functions beginning January 1, 2027.
- Why it matters
- Organizations handling chatbot prompts and outputs should classify those records as regulated medical information where the statute applies, align retention and access controls with CMIA obligations, and preserve audit evidence showing who reviewed clinical recommendations. Discovery and investigation teams will need to distinguish clinical decision support from carved-out scheduling, payment, and informational functions.
- Watch
- The October 29 analysis concerns a law that takes effect January 1, 2027. Applicability depends on the statutory definitions and the function performed; the cited practitioner summary is not a substitute for the enacted text or matter-specific advice.
Sourceshunton.comlegiscan.com
Briefing item
OpenAI challenges a British Columbia privacy investigation
OpenAI has petitioned the British Columbia Supreme Court to quash or narrow a privacy regulator's order seeking answers and documents about its use of personal information.
- What happened
- OpenAI has petitioned the British Columbia Supreme Court to quash or narrow a privacy regulator's order seeking answers and documents about its use of personal information.
- Why it matters
- The dispute places model inputs, derived information, safety review, personalization, moderation, and age-assurance records within a live regulatory evidence demand. AI providers and enterprise users should map where those records reside, who controls them, how long they persist, and what audit trail supports decisions to retain, restrict, disclose, or act on them.
- Watch
- The petition states OpenAI's allegations and requests for judicial review; the court has not ruled on them. The reporting says neither the regulator's full demand nor the complete investigation record was public, limiting independent assessment of scope and proportionality.
Sourcesctvnews.ca
Educational only. Not legal advice. Review sources and matter-specific obligations before relying on any workflow decision.