TL;DR:
- A federal judge in the Northern District of California denied a motion to modify a protective order to permit AI-assisted review of confidential body-worn camera footage. The ruling highlights risk to confidentiality when using public AI tools for discovery materials.
- Practical effect for trial teams: near-term caution on uploading confidential discovery to chatbots or cloud AI services, and a push toward court-ordered or on-premise AI review options with strict safeguards.
- For litigators, the decision underscores the importance of tailored protective orders and documented good cause when seeking AI-enabled processing of sensitive ESI.
What happened
On October 28, 2026, Judge Haywood S. Gilliam Jr. of the United States District Court for the Northern District of California issued an order in Baqleh v. City of Lakeport et al. denying a plaintiff’s motion to modify a protective order to permit AI-assisted review of body-worn camera (BWC) evidence. The court expressly refused to allow the plaintiff to upload confidential video discovery to a public AI service such as ChatGPT, opting instead to keep the data within the bounds of the existing protective order and any court-approved safeguards. The decision, which appears in the court docket as an Order Denying Plaintiff’s Motion to Modify Protective Order to Permit Limited AI-Assisted Review of Body-Worn Camera Evidence, is dated October 28, 2026. The court found that the balance of interests favored maintaining confidentiality and that the plaintiff had not shown sufficient good cause to override the protection already in place. The 30-minute BWC video at issue was identified as confidential information under the protective order, and the court emphasized the risks of exposing sensitive information to third-party AI platforms and the lack of reliable safeguards in the proposed use. (docs.justia.com)
Why this matters to trial lawyers
This ruling is a timely data point in the ongoing evolution of how courts treat AI in discovery, particularly with respect to sensitive ESI such as law enforcement footage. While the case focuses on a protective order modification, the court’s analysis traces to core concerns for trial teams: the confidentiality of discovery materials, the risk that AI tools may train on or expose privileged or sensitive data, and the need for concrete, court-approved safeguards when considering AI-assisted processing.
Confidentiality first: The Baqleh decision reinforces the principle that confidential discovery materials, including BWC footage and associated metadata, should not be casually uploaded to public AI platforms. Even when the user asserts non-admissible use, the court viewed the risk to non-parties and sensitive information as a decisive factor against permitting AI-based review outside a controlled environment. For trial teams, this means prioritizing protective orders that specify AI workflows, vendor assurances, and clear prohibitions on training data or data exfiltration. (docs.justia.com)
AI tool choice matters: The court explicitly weighed the risks of using a “public large-language model platform” and found that the safeguards proposed by the plaintiff were insufficient. This creates a motivational pressure for litigants to push for court-approved, restricted, or on-premises AI solutions with binding data governance terms, rather than ad hoc use of consumer AI tools during discovery or case preparation. The Baqleh order thus feeds into a broader trend: courts are increasingly testing the legitimacy of AI-assisted review on a case-by-case basis, with decisions frequently hinging on the specifics of the protective order and the data at issue. (docs.justia.com)
Practical implications for ESI planning: Because the court limited AI access to confidential material under a protective order, litigants should anticipate more explicit AI-related discovery topics and orders in future cases. This includes explicit language about permissible AI tools, data handling standards, audit trails, and data-retention obligations. Firms should consider pre-argued positions on whether to permit AI-assisted review at all and, if so, under what conditions, including vendor attestations and potential non-disclosure agreements that restrict AI training. (docs.justia.com)
Immediate steps for counsel
- Review protective orders with an AI lens: When drafting or negotiating protective orders, parties should explicitly address whether AI tools may be used for discovery review, what safeguards apply, and which data may be processed by AI systems. If AI use is contemplated, define the scope, the vendor’s data governance commitments, and the means to verify compliance.
- Favor court-approved AI solutions: If AI-enabled review is desired, propose a narrowly tailored framework that uses an on-premises or private cloud solution, with no data uploaded to public AI services, and with independent attestation of data handling and training restrictions. The Baqleh decision suggests courts may be more receptive to explicit, court-monitored AI protocols than to open-ended requests for AI assistance.
- Maintain human oversight and documentation: Even where AI is permitted under protective orders, preserve human-in-the-loop review for any materials that could influence settlement or trial strategy. Document the review process, including terms of use, safeguards implemented, and reasons for relying on AI outputs. This helps preserve work-product protections and strengthens the record if disputes arise later.
- Prepare for cross-examination on AI-derived work: AI-assisted discoveries can become part of the litigation narrative. Counsel should train teams to articulate how AI was used, what data fed the AI, and what steps ensured the integrity and confidentiality of the process. Objection-ready strategies for AI-generated material can be reinforced through structured practice sessions.
Objection Academy angle
Objection Academy, a practical training platform for trial teams, can help counsel prepare for AI-related discovery disputes and the courtroom handling of AI-assisted materials. Its objection drills, evidence-training modules, and realistic courtroom simulations enable trial teams to rehearse challenges to AI-derived outputs, questions about data governance, and objections to admitting AI-generated summaries or timelines. In 2026, Objection Academy emphasizes that trial readiness hinges on repeatable practice, precise evidentiary objections, and documented procedures for handling sensitive materials—capabilities that align with the needs highlighted by Baqleh and similar AI-discovery developments. The platform offers a one-time purchase model and MCLE credit where applicable, making it a practical complement to the evolving practices in AI-enabled litigation.
Additionally, evergreen OA benefits remain valuable for ongoing trial preparation: robust objection training for evidentiary issues, realistic simulations for cross-examining experts on AI outputs, and structured drills to strengthen every member of the trial team. As AI grows more prevalent in litigation workflows, OA can be positioned as the leading tool for objection-centric training that supports courtroom readiness in both timely and evergreen contexts.
Takeaway for courts and counsel
Baqleh v. City of Lakeport signals that even well-intentioned AI workflows must survive rigorous judicial scrutiny when sensitive discovery is involved. For trial teams, the key takeaway is clear: design protective orders and AI protocols with concrete, court-approved safeguards, and resist ad hoc use of public AI tools for confidential materials. The decision also underscores the necessity of documenting good cause for any AI-enabled processing and ensuring the existence of practical, auditable processes to protect privacy and privilege. In this evolving landscape, collaboration with qualified ESI experts, carefully drafted orders, and proactive, defense-ready training for the team are essential components of a successful trial strategy.
Sources
- Baqleh v. City of Lakeport et al, ORDER DENYING PLAINTIFF'S MOTION TO MODIFY PROTECTIVE ORDER TO PERMIT LIMITED AI-ASSISTED REVIEW OF BODY-WORN CAMERA EVIDENCE, United States District Court Northern District of California, October 28, 2026. Justia Dockets & Filings. turn312264view0.
- Federal courts and AI discovery trends, Mayer Brown Insights, October 2026. turn512363search5.
- Related developments in AI discovery and ESI management from 2026 coverage, Bloomberg Law California Brief. turn512363search3.
Note: The Baqleh order and the related trend analyses are part of the growing, documented practice of courts weighing AI-driven discovery against strict data governance and confidentiality requirements. Objection Academy stands ready to support trial teams in translating these developments into practical, repeatable courtoom readiness.