Over decades, technology has repeatedly rewritten the rules of litigation discovery; AI and ephemeral messaging are only the latest frontiers. Modern businesses depend on real time, efficient communications and analysis, across platforms like Signal, WhatsApp and enterprise LLMs such as Microsoft Copilot and Claude. But litigation threatens to turn informal, on-the-fly communications and brainstorming into a major source of discovery battles. The technology surrounding these platforms creates a double-edged threat – if you set your communications to auto-delete, you risk crippling spoliation sanctions; if you save your records and logs, your off-the-cuff brainstorm becomes a potential waiver of privilege or catastrophic admission.
Having a plan before litigation arises is essential to protect your business from these threats.
Threat #1: The Accidental Privilege Waiver via AI Prompts
AI’s speed and ease of use creates numerous opportunities for business leaders to seek greater efficiencies in performing their tasks. It’s not hard to imagine a busy executive pasting sensitive legal advice, internal risk assessments, or contract drafts into a commercial AI tool to “summarize key points” or “draft a response.” But recent developments in the case law make it critical for your employees to understand that inputting protected attorney-client communication or work product into third-party AI platforms can destroy privilege. Even with enterprise LLMs, where you have an expectation of privacy against the information’s use for training or data harvesting purposes, that’s not a guarantee the information remains privileged against discovery in litigation. While attorneys using enterprise AI at their direct instruction may still argue for work-product protection, the law in this area remains unsettled. For unassisted employee communications, prompt logs carry a high risk of exposure. Opposing counsel will subpoena enterprise AI logs specifically seeking raw user prompts. Without privilege, those unvarnished prompts can be introduced as direct admissions.
Threat #2: Ephemeral Messaging and the Spoliation Sanction
Modern businesses also rely on auto-deleting messaging apps (like Signal) for operational speed, security, and privacy. But while that auto-delete function can protect your business against hacks and data theft, once litigation is reasonably anticipated, that protection creates a substantial risk. The duty to preserve potentially relevant materials attaches across all platforms, even casual forms of communications. Failing to turn off auto-delete functions after a duty to preserve arises is a leading trigger for Rule 37(e) spoliation sanctions (and their state court equivalents). Even when the failure to turn off auto-delete is the result of inadvertence, lack of tech fluency, or forgetting to apply the preservation to personal devices used for work (BYOD), the result can be a devastating adverse inference instruction; a judge telling the jury they can assume the deleted messages contained damaging evidence can wreak havoc on your case.
Strategic Action Plan for General Counsels
General Counsels can take a few specific steps before a complaint is ever filed to lower these risks dramatically:
- Establish Enterprise AI Usage Guidelines: Establish explicit corporate governance on what data can and cannot be fed into LLMs. Remember, your AI platform is not an attorney, cannot provide you with legal advice, and cannot guarantee the privacy of your communications in the event of litigation. To ensure that when AI platforms are used for legal research (either by or at the direction of attorneys) the results remain confidential, make sure your enterprise licenses contain strict no-training and confidentiality guarantees.
- Issue Targeted Legal Holds Early: A standard hold memo isn’t enough. Explicitly mandate the manual disabling of auto-delete settings on Signal, WhatsApp, or other self-deleting/ephemeral messaging apps. Make sure that officers and employees apply this not just to their work-issued devices, but to any personal devices used for business.
- Training on AI Prompts: Train your employees, from the C-suite leadership on down, that AI prompts are discoverable text—not private thoughts.
Conclusion
At trial, cases are won by presenting the most compelling narrative. By addressing digital discovery risks today, General Counsels protect their privilege, safeguard their trial posture, and ensure their trial attorneys can put forward the strongest possible defense.