Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Digital Evidence Ai topic
No spam. Unsubscribe anytime.
Attorneys warn of deepfakes and discovery risks as AI enters divorce courts
Summary
Divorce attorney Ian Steinberg told Today's Verdict that AI helps process discovery but also creates risks—deepfakes, fake messages and discoverability of material entered into open AI systems—urging clients to avoid putting sensitive content into public models.
Get email alerts on the Digital Evidence Ai topic
No spam. Unsubscribe anytime.
David Lesh introduced a segment on how digital evidence and artificial intelligence are changing divorce and family-law practice and invited divorce attorney Ian Steinberg to explain current challenges.
Ian Steinberg said AI can be useful for attorneys handling large financial disclosures and lengthy deposition transcripts, but he warned that AI tools are being misused to fabricate evidence in custody and family disputes. "There are fake voice recordings, there are all of these sort of temptations...fake videos," Steinberg said, describing instances in which manipulated content was used to mislead families or courts.
On verification, Steinberg recommended a two-step approach: ask clients whether the content is authentic and then engage digital-forensics experts who specialize in detecting altered images, audio and video. He recounted presentations at legal conferences showing how experts analyze and challenge manipulated material.
Steinberg also addressed discovery and open AI systems, saying recent case law treats material input into open AI systems as potentially discoverable and that attorneys have subpoenaed such records. "Anything put into open AI systems is discoverable," he said on the program, and advised clients not to place privileged or sensitive material into public AI platforms.
Finally, Steinberg urged caution: law firms using closed AI systems can gain efficiencies, but attorneys must avoid relying on AI to create or support legal propositions without verifying sources. He recommended that clients assume anything they post or generate could be seen by a judge and to consult experts when authenticity is disputed.

