EDRM BLOG
Five Faces of the Black Box: How AI ‘Thinks’ and Makes Decisions
Ralph Losey goes deep into how LLM’s work, with visuals to represent how hallucinations might happen. Written for a legal tech audience, this explanation and analysis is applicable to other disciplines.
Does Disclosure of Litigation Hold Directive to Preserve “Texts” Waive Privilege?
A recent decision examines whether identifying “texts” in a litigation hold waives privilege and what it takes to compel disclosure of hold notices.
When the Agent Goes Off-Script: Meta’s AI-Triggered Data Exposure Revives Old Security Fears
Meta’s March 2026 AI agent incident exposed sensitive internal data and highlighted a growing enterprise risk: autonomous systems acting beyond governance controls. For cybersecurity, legal, and eDiscovery professionals, the event signals a shift in how...
Nonsensical Spellings and Fabricated Authority Signal Improper Use of Artificial Intelligence
A federal court found that nonsensical spellings and fabricated authority revealed improper AI use, resulting in dismissal without leave to amend.
HaystackID CoreFlex Wins 2026 Artificial Intelligence Excellence Award for Advancing GenAI-Driven Legal Workflows
HaystackID’s CoreFlex platform earns 2026 AI Excellence Award, highlighting its role in transforming legal data workflows with generative AI and automation.
The M&A Risk of Confusing Market Velocity with Marketing Capability
This analysis examines a critical M&A risk: mistaking market-driven momentum for durable marketing capability. It outlines how this misread impacts valuation, diligence, and integration outcomes.
Well-Reasoned “Hallucination” Analysis
A federal court decision in Brownfield v. Cherokee Co. School Dist. clarifies that failing to verify AI-generated citations can trigger Rule 11 sanctions, even for pro se litigants.
Weekly Letter to Our EDRM Global Community – 24 March 2026
This week’s EDRM community update features recent blog posts, upcoming webinars, notable podcasts, and key announcements. Stay engaged with the EDRM community for the latest insights and support.
Evidence Sufficient to Demonstrate that Audio Recording Was Not a Deepfake
In Burnley v. Valentin (E.D. Va. 2026), the court held that an audio recording was properly authenticated despite a deepfake challenge, emphasizing FRE 901’s low threshold, witness testimony, and voice identification.
Defensible by Design: What Legal Teams Must Get Right About AI Privilege Workflows
AI is now operational in privilege review, but defensibility remains the standard. Drawing on a Legalweek 2026 panel, this article explores validation protocols, the distinction between classification and logging, and the importance of Rule 502(d)...
Plaintiff Sold Her Cell Phone After Litigation Commenced
In Hernandez v. DiBiaso (N.D. Ill. Mar. 9, 2026), the court held that selling a cell phone after litigation began violated preservation duties under Rule 37(e). Although the loss of ESI caused prejudice, the court...
Litigation Strategy Moves at the Speed of Insight
When litigation hits, insight can’t wait. This article explores how AI-powered case strategy helps agencies quickly understand complex records, assess risk, and act sooner, without compromising security or defensibility.