A photograph of the crash scene. Surveillance footage. A recording. A text message. A video showing what happened in the moments before an injury.

There is something inherently persuasive about being able to tell a jury: You don’t have to take my word for it. Look for yourself.

Artificial intelligence is beginning to complicate that assumption.

Today, readily available technology can create realistic photographs of events that never happened, clone a person’s voice to say words they never spoke, alter existing video, and generate entirely fictional footage that may appear authentic to the human eye.

For trial lawyers, that creates an entirely new evidentiary problem: What happens when seeing is no longer believing?

Deepfakes Are No Longer Science Fiction

The term “deepfake” originally referred primarily to manipulated videos, but today the concept is much broader. AI can be used to fabricate or manipulate photographs, videos, voices, social-media content and other forms of digital evidence. There are sophisticated creations which are distinguishable from “cheap fakes” made with less advanced technology and “shallow fakes,” where genuine content is manipulated—for example, by changing its speed or taking it out of context.

Now, creating convincing material no longer necessarily requires sophisticated technical knowledge.

A deepfake can be created using only a couple photographs and clips of someone’s recorded voice. The generated deepfake can be so convincing it can deceive family members. That should get the attention of anyone involved in litigation.

Imagine receiving an audio recording supposedly capturing a driver admitting he was texting immediately before a collision.

Or surveillance footage appearing to show an injured plaintiff lifting something heavy.

Or a photograph supposedly showing the condition of a property immediately before someone fell.

Five years ago, the primary question might have been: What does the evidence show?

Now it may be: Is this evidence even real? Is it AI evidence?

Personal Injury Cases May Be Particularly Vulnerable

Modern personal injury cases are increasingly built around digital evidence.

Dash cameras, surveillance systems, Ring cameras, cell-phone photographs, text messages, social-media posts, vehicle data and electronically stored information can become critical pieces of a case.

AI creates risks on both sides of the courtroom.

A dishonest litigant could theoretically manufacture evidence supporting a claim. But fabricated evidence could just as easily be used against an injured person—a supposed social-media photograph, altered surveillance video or fabricated recording could create an entirely false picture of someone’s physical abilities, activities or statements.

Lawyers now need to begin thinking differently about digital evidence from the moment a case begins.

The Original File May Matter More Than Ever

A screenshot is not necessarily the same thing as the underlying evidence.

Experts identify several warning signs lawyers should consider when evaluating suspicious digital evidence: evidence that seems almost too perfect, a poor or unexplained chain of custody, an inability to produce the original file or device, and the absence of native files and associated metadata.

That makes preservation increasingly important.

If a photograph matters, preserve the original photograph.

If a video matters, preserve the original video file.

If something was recorded on a phone, preserve the phone.

If surveillance footage matters, obtaining the original footage directly from the source may become much more important than relying upon a forwarded copy, screenshot or screen recording.

The history of a digital file—where it came from, who possessed it, how it was transferred, and whether its underlying metadata remains intact—may eventually become almost as important as what appears on the screen.

Courts Are Already Wrestling With the Problem

In United States v. Bell, video from a Ring surveillance system contained unexplained gaps, including one exceeding 22 minutes. The person whose system generated the footage was unavailable to testify, and questions existed about how the video had been obtained and transferred. The defendants argued that the footage had been tampered with.

The appellate court in Bell ultimately concluded there was sufficient authentication and treated the gaps as an issue affecting the weight of the evidence rather than its admissibility. But a dissent raised a much larger concern: traditional methods for authenticating video evidence were developed decades before AI-generated video and modern deepfakes became possible. That tension is likely to become increasingly important.

Historically, disputes about photographs and videos frequently centered on whether the evidence fairly depicted what it purported to show.

AI introduces another possibility altogether: the depicted event may never have happened.

Even AI May Not Be Able to Tell Us What AI Created

Perhaps the most surprising part of this issue is the limitation of today’s deepfake-detection technology.

Experts explain there are many different AI detection tools. However, even when using these AI detection tools, different conclusions can be reached. Some AI detection tools will identify a piece of information as AI generated, others identify it as human-generated, and still others cannot decide.

In other words, asking another AI program whether something was created by AI is not necessarily going to solve the problem.

Instead, determining authenticity may require looking at the entire body of evidence: the original device, metadata, chain of custody, surrounding circumstances, inconsistencies within the recording, and potentially testimony from qualified experts.

Why Lawyers May Have to Address Authenticity Much Earlier

Waiting until trial to challenge suspicious evidence may be too late.

Questions concerning AI-generated evidence need to be identified during discovery rather than on the eve of trial. Resolving an authenticity dispute could require examination of the original device, additional discovery, expert analysis, protective orders and even evidentiary hearings.

And if the original device suddenly disappears, the dispute can transform from an authentication problem into a preservation or spoliation problem. Metadata may also be critical—and may disappear depending upon how a file is copied, transmitted or generated.

For personal injury lawyers, that means evidence preservation cannot be an afterthought.

The Other Danger: The “Deepfake Defense”

There is another side to this problem that may ultimately be just as significant.

As fake evidence becomes easier to create, genuine evidence becomes easier to deny.

A person confronted with an authentic recording can simply say: “That’s AI. I never said that.”

Someone shown in an unfavorable video can argue: “That’s a deepfake. I was never there.”

This phenomenon has sometimes been called the liar’s dividend: the existence of convincing artificial evidence gives people a new way to attack genuine evidence.

That means lawyers will increasingly need to be prepared not only to expose fake evidence, but also to prove that real evidence is real.

If you’re a lawyer concerned about AI evidence, the Sedona Conference has a great digital asset which provides more information to guide your analysis which can be found online here.

If you’re a potential client and would like to read more about my firm, check out my profile here.

What Should You Do With Important Digital Evidence?

For an injured person, the practical lesson is surprisingly simple:

Don’t alter the evidence. Preserve it.

Don’t crop an important photograph and delete the original. Don’t edit a video before sending it to your lawyer. Don’t rely solely upon screenshots of important messages. Don’t discard an old phone simply because you purchased a new one.

And certainly don’t use artificial intelligence to “improve,” reconstruct or recreate evidence.

Give your attorney the original material whenever possible and explain exactly where it came from.

The technology surrounding evidence may be changing rapidly. The fundamentals of a strong case are not.

Preserve the evidence. Document where it came from. Investigate it carefully. And never assume that something is true simply because it looks real.

At Salter Law, we believe serious personal injury cases require more than collecting evidence—they require understanding it. As technology changes the way photographs, videos, recordings and electronic information are created and challenged, careful investigation and thoughtful litigation strategy matter more than ever.

If you’ve been seriously injured in Louisiana and have questions about the evidence surrounding your accident, contact Salter Law to discuss your case.