In July 2025, a Maine police department triggered public outrage when it used artificial intelligence (AI) enhanced image editing to insert a department patch into a photo of seized drug evidence. The application inadvertently altered other parts of the image as well, including text and packaging graphics. The modifications were spotted by members of the public when the altered image was uploaded to the agency’s social media.1
In January 2026, a Texas police department was criticized when it used AI enhanced image editing software to sharpen a blurry mugshot of a retail theft suspect.2 While the scope of the image editing seems innocent on its surface, the legal community quickly expressed concern that any manipulation of photographic evidence by law enforcement could undermine its credibility in legal proceedings.3
Regardless of whether you are a proponent or critic, AI is finding its way into many software applications used for day-to-day business and life activities. Along with the benefits being promoted, users should also be aware of the risks posed by this emerging technology. One concern is that the AI can make factual errors known as “hallucinations.” A recent personal experience I had involved posing a question to an AI application involving Pa. boiler code. That inquiry produced answers several of which were found to have been extracted from the Pa. elevator code! Another concern is that this technology makes it easy to manipulate images, video and voice. In July 2025, a Florida mother was conned out of $15,000 following a phone call with whom she thought was her daughter. She was actually interacting with an AI-generated “voice clone.” The mother reported that someone who sounded exactly like her daughter called pleading for her to send money for bail. This was all fake – the daughter had not been arrested, nor had she called her mother from jail. Investigators believe the fraudster created a voice clone of the daughter using audio pulled from unsecured social media.4
For these reasons and others, the Delaware Valley Trusts recommend that public employers adopt a basic AI use policy covering their organization. This can be used to establish ground rules and behavior expectations for employees, volunteers and elected officials. From a proactive risk management perspective and given the potential for significant mayhem, it makes sense to determine potential approved uses of AI as well as establish prudent controls on how AI may be used in the organization.
To help with this emerging risk, DVT offers a sample AI use policy. This policy is a good starting point for internal discussions. As with most legal matters, we encourage members to review the sample policy with their solicitor prior to formal adoption. DVT’s staff also stands ready to provide members with AI-related risk management support. DVPLT members with questions or concerns are encouraged to contact their lead Risk Control Consultant (lwolf@dvtrusts.com, lconway@dvtrusts.com or mmedori@dvtrusts.com) or call the Risk Control Department at 215-706-0101.
Sources:
1Allen, Bailey. “Maine Police Department Apologizes after Posting AI-Altered Evidence Photo on Social Media.” Portsmouth Herald, Portsmouth Herald, 3 July 2025, www.seacoastonline.com/story/news/technology/2025/07/03/westbrook-maine-police-digital-altered-photo-evidence-social-media-apology/84456838007/.
2“Texas Police Review AI Use after Edited Mugshot Draws Scrutiny.” Police1, Police1, 16 Jan. 2026, www.police1.com/law-enforcement-policies/texas-police-review-ai-use-after-edited-mugshot-draws-scrutiny?utm_source=Sailthru&utm_medium=email&utm_campaign=P1-RollCall-1-16-26&utm_term=P1+Roll+Call.
3Grindley, Stephanie. “‘I Would Have a Field Day’: Criminal Defense Attorney Speaks on Westbrook AI Photo Scandal.” WGME, 4 July 2025, wgme.com/news/i-team/i-would-have-a-field-day-criminal-defense-attorney-speaks-on-westbrook-ai-photo-scandal-maine-evidence-drugs.
4Egozi, Gabriella. “Florida Woman Swindled out of $15K after Ai Clones Daughter’s Voice.” NBC 6 South Florida, NBC 6 South Florida, 17 July 2025, www.nbcmiami.com/news/local/florida-woman-swindled-into-giving-15k-after-ai-clones-daughters-voice/3660507/.