Evidence authentication is not a new problem in state courts. But the rise of artificial intelligence has made it easier to enhance, alter, or create fake digital evidence that looks convincingly real, while making detection more difficult than ever. The mere existence of AI technology now casts doubt on the authenticity of all digital evidence, threatening public trust in the justice system.
The TRI/NCSC AI Policy Consortium for Law & Courts, a partnership between the national Cetner for State Courts and the Thomson Reuters Institute, has released the second edition of “Evidentiary Issues Raised by Artificial Intelligence,” updated to help court leaders and judges understand how AI is reshaping traditional approaches to using and authenticating digital evidence.
The rapid advancement of deepfake technology has created a technological arms race between creation and detection methods. The expanded use of acknowledged AI-generated evidence has also changed how we think about accuracy and authentication. The report identifies two general approaches for courts to address this challenge.
1. Increase access to technologies and technical experts that can detect deepfakes. The report notes that technological solutions should complement, rather than replace, legal frameworks for evaluating evidence.
2. Update court rules and procedures. Several states have introduced rules addressing both acknowledged and unacknowledged AI-generated evidence. Utah’s Rule of Evidence 707, for example, requires machine-generated evidence offered without an expert witness to satisfy the same reliability standards as expert testimony. Other states addressing AI in court rules — which could apply to both deepfake technologies and accuracy issues related to acknowledged AI — include Connecticut, Florida, Louisiana, New York, and Oklahoma.
Since the first edition of this report, court cases addressing specific instances of deepfakes have emerged. A California court dismissed a civil case after determining that plaintiffs had submitted deepfake videos and altered images. A New York court reversed a family court finding of child abuse, holding that video evidence had not been properly authenticated.
The report also offers guidance on what judges can do today, including questions to ask about the source and authenticity of digital evidence.
The consortium’s work supports NCSC’s Strategic Agenda priorities, which include recognizing the significant impact artificial intelligence and other technological innovations could have on court operations and working with partners to provide advice and guidance on the best use of these tools and the ethical and practical considerations they inevitably raise.
To explore the report’s full findings and guidance, along with other digital evidence resources for judges, visit www.ncsc.org/resources-courts/ai-generated-evidence-guide-judges.
The TRI/NCSC AI Policy Consortium for Law & Courts, a partnership between the national Cetner for State Courts and the Thomson Reuters Institute, has released the second edition of “Evidentiary Issues Raised by Artificial Intelligence,” updated to help court leaders and judges understand how AI is reshaping traditional approaches to using and authenticating digital evidence.
The rapid advancement of deepfake technology has created a technological arms race between creation and detection methods. The expanded use of acknowledged AI-generated evidence has also changed how we think about accuracy and authentication. The report identifies two general approaches for courts to address this challenge.
1. Increase access to technologies and technical experts that can detect deepfakes. The report notes that technological solutions should complement, rather than replace, legal frameworks for evaluating evidence.
2. Update court rules and procedures. Several states have introduced rules addressing both acknowledged and unacknowledged AI-generated evidence. Utah’s Rule of Evidence 707, for example, requires machine-generated evidence offered without an expert witness to satisfy the same reliability standards as expert testimony. Other states addressing AI in court rules — which could apply to both deepfake technologies and accuracy issues related to acknowledged AI — include Connecticut, Florida, Louisiana, New York, and Oklahoma.
Since the first edition of this report, court cases addressing specific instances of deepfakes have emerged. A California court dismissed a civil case after determining that plaintiffs had submitted deepfake videos and altered images. A New York court reversed a family court finding of child abuse, holding that video evidence had not been properly authenticated.
The report also offers guidance on what judges can do today, including questions to ask about the source and authenticity of digital evidence.
The consortium’s work supports NCSC’s Strategic Agenda priorities, which include recognizing the significant impact artificial intelligence and other technological innovations could have on court operations and working with partners to provide advice and guidance on the best use of these tools and the ethical and practical considerations they inevitably raise.
To explore the report’s full findings and guidance, along with other digital evidence resources for judges, visit www.ncsc.org/resources-courts/ai-generated-evidence-guide-judges.
