Lead story
Connecticut Magistrate Rules AI Prompts Are Discoverable in Expert Methodology
Magistrate Judge Thomas O. Farrish (D. Conn.) ordered the Conservation Law Foundation to disclose AI prompts used by expert Dr. Naomi Oreskes in Conservation Law Foundation v. Shell Oil Co., ruling they form part of the expert's methodology under FRCP 26(b). The court rejected CLF's argument that a Rule 29 agreement shielding 'expert notes, drafts, or communications' covered AI prompts.
- What happened
- Magistrate Judge Thomas O. Farrish (D. Conn.) ordered the Conservation Law Foundation to disclose AI prompts used by expert Dr. Naomi Oreskes in Conservation Law Foundation v. Shell Oil Co., ruling they form part of the expert's methodology under FRCP 26(b). The court rejected CLF's argument that a Rule 29 agreement shielding 'expert notes, drafts, or communications' covered AI prompts.
- Why it matters
- Legal teams using AI tools in expert analysis must now document and be prepared to produce AI prompts as part of methodology disclosure. The ruling signals that courts will treat AI prompt engineering as integral to expert testimony reliability, adding a new layer to expert report preparation and discovery obligations.
- Watch
- Ruling issued by a magistrate judge, not a district judge or appellate court, so precedential authority may be limited. Different courts may reach different conclusions on AI prompt discoverability.
Sourcesreason.comlegaltechdigest.comecf.ctd.uscourts.gov
Briefing item
DOJ Antitrust Division's Reported AI Use Raises eDiscovery Bar for HSR Responders
A May 13 MLex report cited an Antitrust Division official confirming the DOJ is using AI to detect and investigate anticompetitive conduct, including algorithmic pricing and competitor information-sharing. The disclosure raises the bar for companies facing Hart-Scott-Rodino Second Requests, civil investigative demands, or grand-jury subpoenas.
- What happened
- A May 13 MLex report cited an Antitrust Division official confirming the DOJ is using AI to detect and investigate anticompetitive conduct, including algorithmic pricing and competitor information-sharing. The disclosure raises the bar for companies facing Hart-Scott-Rodino Second Requests, civil investigative demands, or grand-jury subpoenas.
- Why it matters
- Companies facing antitrust investigations must assume the government is using AI-assisted review to detect patterns. Preservation, validation, and production workflows need to account for machine-assisted government scrutiny of pricing data, chat transcripts, and model artifacts.
- Watch
- The DOJ's AI use was reported by MLex citing an official; formal public documentation of methodology has not been disclosed. The RealPage consent judgment remains under Tunney Act review.
Sourcesedrm.netcomplexdiscovery.com
Briefing item
Preservation Order Denied Under 'Cry Wolf' Doctrine in Zeta Global Data Privacy Litigation
In In Re Zeta Global Data Privacy Litigation, 2026 WL 1283618 (S.D.N.Y. May 11, 2026), the court denied plaintiffs' request for a preservation order, holding that defendants' assurances of compliance with preservation duties were appropriately responsive and that 'mere speculative assertions' of spoliation risk are insufficient to merit a court order.
- What happened
- In In Re Zeta Global Data Privacy Litigation, 2026 WL 1283618 (S.D.N.Y. May 11, 2026), the court denied plaintiffs' request for a preservation order, holding that defendants' assurances of compliance with preservation duties were appropriately responsive and that 'mere speculative assertions' of spoliation risk are insufficient to merit a court order.
- Why it matters
- Parties seeking preservation orders must present a factual predicate showing actual risk of destruction, not mere dissatisfaction with the opposing party's discovery responsiveness. Defendants who provide specific preservation details and demonstrate compliance can defeat preservation order motions even in high-stakes class actions.
- Watch
- Source-date: Article published on EDRM May 20, 2026; ruling issued May 11, 2026. Analysis based on author Michael D. Berman's summary of the opinion.
Sourcesedrm.netediscoveryllc.com
Briefing item
Anthropic's Claude AI Expands Legal Aid Access with 20+ Connectors and 12 Legal Plugins
Anthropic launched 20+ Model Context Protocol connectors and 12 legal-domain plugins for Claude AI in May 2026, integrating with Westlaw, DocuSign, Everlaw, and other legal platforms. The Justice Technology Association partnered with Anthropic to embed Claude in access-to-justice platforms, and LawDroid released a free Legal Aid Plugin with 15 skills for civil legal aid programs.
- What happened
- Anthropic launched 20+ Model Context Protocol connectors and 12 legal-domain plugins for Claude AI in May 2026, integrating with Westlaw, DocuSign, Everlaw, and other legal platforms. The Justice Technology Association partnered with Anthropic to embed Claude in access-to-justice platforms, and LawDroid released a free Legal Aid Plugin with 15 skills for civil legal aid programs.
- Why it matters
- The expansion of Claude AI into legal aid and civil legal services represents a significant shift in how AI tools are deployed beyond commercial law firms. Legal professionals should monitor integration capabilities with MCP connectors and assess how plugin ecosystems may affect workflow design and service delivery models.
- Watch
- Data measuring Claude AI's impact on legal aid outcomes and access-to-justice metrics remains limited. Impact on actual justice accessibility requires further study.
Sourceslegaltechdigest.comlawnext.comclaudeapi.com
Briefing item
Relativity Adds Claude Compliance API Integration for Enterprise Data Collection
Relativity announced integration with Anthropic's Claude Compliance API to enable collection of Claude Enterprise data within the RelativityOne platform. The integration allows organizations to collect, review, and produce communications and data generated through Claude Enterprise as part of eDiscovery workflows.
- What happened
- Relativity announced integration with Anthropic's Claude Compliance API to enable collection of Claude Enterprise data within the RelativityOne platform. The integration allows organizations to collect, review, and produce communications and data generated through Claude Enterprise as part of eDiscovery workflows.
- Why it matters
- As AI prompt discoverability rulings emerge (see Story 1), organizations need practical tools to collect and review AI-generated content. This integration provides a vendor-supported path for incorporating Claude Enterprise data into defensible eDiscovery workflows, reducing the gap between legal obligation and technical capability.
- Watch
- Source-date: Article published on EDRM May 19, 2026. Technical details limited to vendor announcement; independent validation of collection completeness not yet available.
Sourcesedrm.netrelativity.com
Educational only. Not legal advice. Review sources and matter-specific obligations before relying on any workflow decision.