Gids voor de samenleving
Discovery of AI Prompts and Chat Logs as Evidence
AI prompts and outputs may be relevant electronically stored information, but discovery remains subject to relevance, proportionality, privilege, work-product protection, and case-specific orders.
Op deze pagina3 minuten lezen
Overzicht
A 2025 OpenAI log-preservation order was later terminated for future data, with limited historical and identified-account exceptions.
Diepe duik
Prompts, outputs, settings, and related records can be electronically stored information when they are relevant to a claim or defense. Under the current Federal Rules of Civil Procedure, discovery is limited to nonprivileged matters relevant to a claim or defense and proportional to the needs of the case. A prompt is not automatically discoverable merely because it exists, and it is not automatically privileged because a lawyer used an AI tool. Privilege and work-product protection depend on context, purpose, confidentiality, and governing law. Attorney-directed prompts that reveal litigation strategy may present a different question from routine business use or a consumer’s personal chat. A court’s decision is specific to the record, parties, and claims. On May 13, 2025, a magistrate judge in a New York copyright action ordered OpenAI to preserve and segregate certain output-log data. An October 9, 2025 stipulated order terminated the general going-forward obligation effective September 26, 2025, while previously segregated data remained preserved subject to stated exceptions. The stipulation also continued preservation for certain identified-domain accounts. OpenAI’s October 22 update confirmed the general obligation had ended. These were case-specific terms, not a universal order compelling all chats or requiring providers to retain all user data. When a preservation duty may apply, identify relevant systems, custodians, retention settings, exports, attachments, and metadata; suspend deletion only as required by the hold and applicable law. Keep enough version and settings information to authenticate an output when it matters. For discovery requests, assess scope, burden, privacy, relevance, and privilege, and meet and confer about limits and production format. This overview is educational and does not determine what a particular party must preserve or produce.
Strategische impact
Risico en veiligheid
Catastrofale en alledaagse schade door AI hangt af van wie de risico's begrijpt en wie kan handelen.
Duidelijkere beslissingen
Publieke en professionele geletterdheid bepalen of een krachtig veiligheidsbeleid politiek mogelijk is.
Door de hype heen snijden
Duidelijke verklaringen verminderen de kans op hypes, laboratorium-PR en vaag ethisch theater.
The Future of Discovery of AI Prompts and Chat Logs as Evidence
Courts continue to address preservation, discovery, privilege, and evidence questions involving AI records. The OpenAI order shows why later stipulations matter: its general going-forward duty ended in September 2025, while some previously segregated or identified-account data remained subject to preservation. That outcome applies to the litigation, not every provider or user. Organizations should refresh legal-hold guidance against current rules, orders, retention settings, and contracts. Clear tool and export records can support later authentication, while access controls and data minimization limit unnecessary exposure. Preservation obligations can be modified or terminated by later orders, so current status should be checked before describing a litigation hold.
Implementatie in de echte wereld
A company preserves relevant prompts and outputs after a litigation hold covers the workflow.
Counsel separates confidential legal analysis from ordinary product-development chats.
A party proposes scoped search terms and date limits for AI-generated logs.
A team records model version and settings to explain how an evidentiary output was produced.
Risico's en vangrails
Existentieel risico behandelen als sciencefiction, terwijl capaciteiten zich vermenigvuldigen.
De veiligheid van oppervlakteproducten verwarren met uitlijning onder hoge autonomie.
Hierdoor blijven niet-Engelstalige en niet-deskundige doelgroepen alleen bronnen van lage kwaliteit over.
Implementatie routekaart
Afzonderlijke risico's voor productschade, misbruik en verlies van controle/verkeerde uitlijning.
Vraag welk bewijs uw kijk op tijdlijnen en ernst zou veranderen.
Geef de voorkeur aan primaire bronnen en concrete evaluaties boven marketingclaims.
Identificeer één actiepad: carrière, beleid, financiering of vaardigheden – niet alleen bewustwording.
Blijf verkennen
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the Discovery of AI Prompts and Chat Logs as Evidence quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Veelgestelde vragen
What is Discovery of AI Prompts and Chat Logs as Evidence?
AI prompts and outputs may be relevant electronically stored information, but discovery remains subject to relevance, proportionality, privilege, work-product protection, and case-specific orders. A 2025 OpenAI log-preservation order was later terminated for future data, with limited historical and identified-account exceptions.
A litigation hold covers a team’s use of an AI drafting tool. What should the organization identify first?
A hold should identify information sources and records relevant to the matter.
What did the October 9, 2025 stipulation change about the May OpenAI log-preservation order?
The stipulation ended the broad future obligation as of September 26, 2025, but preserved certain previously segregated data and contained specific exceptions.
Does an attorney’s use of an AI tool automatically make the prompt privileged?
Privilege and work product are context-dependent; tool use alone does not decide them.
Which limits apply when a party requests prompts and chat logs in federal civil discovery?
Rule 26 discovery is limited to nonprivileged information relevant and proportional to the case, subject to applicable orders.
Why record the model or tool version and relevant settings for an AI output used as evidence?
Version and settings help explain provenance but do not prove correctness.
Blijf leren
Gerelateerde gidsen
Er zijn meer handleidingen voor dit onderwerp geselecteerd