The deepfake dimension of admissible evidence under Daubert and F.S. 90.702 is an increasingly complex and challenging reality…Here are some expert witness tips…

Photographs, videos, audio recordings and documents are all subject to the Deepfake quagmire.

Besides software that is purported to detect altered images and content, trial lawyers need to be ready with human expertise for testimony from the witness box.

Here is a list of the kind of expert testimony you may need to consider depending upon the needs of your case.

·        Imaging and Photography Experts

·        Audio forensic specialists

·        Digital forensic analysts

·        Forensic Linguists

·        Computer vision and AI Experts

·        Cryptography and Security Experts

Digital experts can scrutinize metadata, how files are built and created…all in an effort to detect deepfakes. In videos, for example, “flicker” (skin texture issues), subtle changes in facial structure, and “lip-sync mismatches” can be identified.

Imaging, video and photography experts can analyze background and foreground lighting and details which AI may not be able to copy. “Vertical cropping” and poor facial and image quality which may indicate a photo of a printed page can possibly be identified. “Skin artifacts” and human facial inconsistencies may be detected as deepfakes. Other lighting inconsistencies can sound alarms. Even hair and eyebrows can produce warning signs for litigators and their experts. Eyes in photographs can signal issues. “Geometric consistency” with lighting and shadows are elements which can be examined. “Contextual weirdness” can be assessed. Depth of field inconsistencies can also produce indicators for greater review. “Noise pattern mismatches” can be signs of trouble.”Camera fingerprints” can show telltale camera and image questionability issues.

Audio experts can analyze vocal biomarkers, breathing patterns and spectral data to detect “synthetic speech.”

Forensic Linguistics can analyze vocabulary, style patterns and syntax.

Computer Vision and AI Experts can locate telling “fingerprints” from AI models used in producing AI generated media.

Cryptography & Security Experts can opine on digital signature and block chain…to verify authentic content.

The information above is based upon careful research from Perplexity, Google Notebook LM and Chat GPT-5. They point out that human expertise is a safeguard for credible testimony in Court and can be used with AI Deepfake software programs.

As with all artificial intelligence products, using software alone to detect false images may not be enough, and can produce inaccurate results and flawed conclusions.

Your safeguard is to retain qualified experts who can explain complex AI analysis to Judges and Jurors.

“Evidence verification” is the key as AI advances…with qualified experts who can meet the demanding (and escalating) AI requirements of Daubert.

For more information on this topic, please contact the Editor.

Deepfake detection for the Courtroom is part of a formula which links experts and software. Daubert admissibility is the goal for AI deepfake testimony.