Notable developments in the e-discovery and information governance space in 2019 included the lead-up to the California Consumer Privacy Act, which came into effect January 1, 2020, and strategies to reduce burdens and costs of complying with discovery orders, including Rule 45 subpoenas. Last year, we also saw artificial intelligence and litigation holds garner attention, as they relate to discovery practices and as they are addressed by courts. Below are a few highlights from 2019.

Artificial Intelligence. The rise of artificial intelligence has opened the door for the technology itself to be the subject of litigation, as well as the technology providing additional forms of discovery.

  • Businesses’ increased reliance on artificial intelligence as an integral aspect of its consumer-facing work has raised concerns about how businesses will be able to defend their technology if their decision-making is challenged in court. To that end, the owner of artificial intelligence should design the model for explainability and defensibility:
    • Explainability means that the tool is transparent (easy to identify important factors in the tool’s operation) and interpretable (easy to identify and explain how the tool weighs those factors and derives them from input data) and has clear data provenance (easy to identify where input originated). These qualities are best achieved if:
      • The company tells the data scientists how it wants the tool to work;
      • The tool stores facts about how it arrived at results in a manner approved by the company’s e-discovery/information governance team; and
      • The company employs people who know how to explain the tool’s results (AI explainers);
      • The company employs people who continually test and modify the tool (AI sustainers).
    • Defensibility is ideally demonstrated by the company’s employees being able to explain the tool’s results—being able to serve as defendant witnesses who can justify decisions they made regarding the development and design of the tool and its processes.
  • Artificial intelligence is also often used by companies to track aspects of its customer service, which can lead to a company having tens of thousands of hours of audio data, which is discoverable electronically stored information (“ESI”) in civil litigation under Federal Rule of Civil Procedure 34. Companies can reduce the costs and burdens associated with producing this kind of data by:
    • Using technology that produces searchable data;
    • Leveraging advanced analytic technology to limit the volume of data to be processed;
    • Narrowing the scope of inquiry by conducting witness interviews and a limited review of written communications before engaging in audio processing;
    • Reviewing and tailoring retention policies to meet applicable regulatory requirements and business needs; and
    • Instituting a collection protocol for audio data.

    Litigation holds. In light of the 2015 amendments to the Federal Rules of Civil Procedure (Rules) and evolving data privacy obligations, companies need to be sure that their litigation hold policies and procedures comply with the Rules and applicable regulations. In addition, litigants have taken a keen interest in opposing parties’ preservation of data, which can lead to costly discovery and, potentially, sanctions.

    • The Sedona Conference issued a new publication in June 2019, Commentary on Legal Holds, Second Edition: The Trigger & The Process, which provides guidance for determining when to issue a litigation hold, what should be covered by the hold, how to implement the hold, and how to address foreign data protection laws or regulations that might affect an organization’s preservation measures outside of the United States. Key takeaways include: 
      • Many of the guidelines are formulated to demonstrate reasonableness and good faith (see Guideline Nos. 2, 3, 4, 5, 6);
      • Companies should not simply issue a litigation hold and call it a day (see Guideline Nos. 9, 10, 11); and
      • Legal holds should include provisions pertaining to releasing the hold (Guideline 11). 
    • Discovery on discovery, which often takes the form of interrogatories and 30(b)(6) deposition notices, is permissible and expensive and can be a prelude to sanctions. In an attempt to avoid discovery on discovery, parties should make early and meaningful disclosures. Courts:
      • Expect that litigants will meet and confer before considering a motion to compel discovery on discovery; 
      • May impose measures such as additional discovery and restoration from alternative sources as alternatives to discovery on discovery; and
      • Often require a party seeking discovery on discovery to demonstrate not only the deficiency of the producing party’s production but also that the deficiency resulted from bad faith.