Authenticating Photographs Obtained from Social Media Platforms

In my last post, I wrote about how a party might authenticate a Facebook direct message or other text-based electronic communication. That post focused on how the proponent of the evidence might establish who wrote the message, i.e., authorship. But what if a party wants to introduce a photograph that was posted on a social media platform? The concept of authorship doesn’t really apply, and in the age of Photoshop and AI-generated images, courts may have serious concerns about the accuracy of online images.

An example from an actual case. Consider United States v. Farrad, 895 F.3d 859 (6th Cir. 2016). The case began when “concerned citizens” notified law enforcement that they had seen Malik Farrad, a convicted felon, carrying a gun. An officer using a fake Facebook account sent a friend request to an account that appeared to belong to Farrad. The account holder accepted the request, and the officer promptly saw on the account a photograph of three guns in a bathroom. This prompted the officer to obtain a search warrant for the Facebook account. He served it on Facebook and received data in response, including several photographs that had recently been posted on the account showing (what appeared to be) Farrad holding (what appeared to be) guns against a backdrop (that appeared to be) Farrad’s apartment.

The officer charged Farrad with being a felon in possession of a firearm. No concerned citizens testified at trial. The photographs were virtually the only evidence against Farrad, bolstered by officers’ testimony that criminals often post pictures of their crimes online, usually contemporaneously. Farrad argued that the photographs couldn’t be authenticated, might not depict real guns, and in any event, were insufficient to convict given the lack of evidence that they were accurate and the lack of evidence regarding when or where they were taken. The trial judge admitted the photographs on the theory that they were Facebook’s business records and a jury convicted Farrad. He appealed, pressing the same basic arguments.

What about the traditional foundation for photographs? Photographs have been used as evidence in court for more than a hundred years, and we have well-established rules concerning the necessary foundation. A photograph may be admitted for illustrative purposes if a witness testifies that it will help illustrate his or her testimony. See, e.g., State v. Little, 253 N.C. App. 159 (2017) (pictures of a motorcycle that the defendant apparently posted on Instagram were properly admitted to illustrate the testimony of the motorcycle’s owner regarding how the motorcycle looked at the time the defendant allegedly stole it). A photograph may be admitted for substantive purposes if a witness testifies that it fairly and accurately depicts a relevant scene known to the witness, or if a witness testifies that the camera was working properly at the time the photograph was taken and the resulting photograph has not been altered before trial.

The government did not use these settled pathways in Farrad, presumably because it had no witness who could testify that the photographs fairly and accurately depicted Farrad holding a gun, nor any witness who could assert that the camera that captured the photographs was working well. Perhaps the photographs could have been admitted to illustrate the officer’s testimony about the course of his investigation, but if admitted on that theory, they would not have been substantive evidence and so could not have been used to support Farrad’s conviction.

This is a relatively common situation in criminal cases: investigators locate a social media account that appears to belong to the defendant, and find seemingly incriminating photographs posted by the account holder. But the investigators don’t know who took the pictures, what camera was used, or even who was present, so there is no way to lay the traditional foundations. Some courts have applied the traditional tests and have excluded photographs under these circumstances. See United States v. Winters, 530 Fed. Appx. 390 (5th Cir. 2013) (unpublished) (photographs found on defendant’s social media accounts that appeared to show guns, “hundreds of thousands of dollars” in cash, and kilograms of cocaine packaged for sale were improperly admitted as no witness could “recognize and identify the objects in the photos”). Cf. State v. Spellman, 234 N.C. App. 667 (2014) (describing trial court’s exclusion, based on lack of authentication, of a photograph posted on what appeared to be the defendant’s Facebook page).

The business records theory. In Farrad, the government tried a new way to authenticate the photographs: it contended that they were Facebook’s business records. Under Federal Rule of Evidence 902(11), certified business records are self-authenticating, and Facebook confirmed that the information it provided came from its records of the account at issue. (North Carolina also provides for the authentication of evidence as business records, but the relevant provision is located in the business records hearsay exception, Rule 803(6).)

As noted above, this theory worked in the trial court. However, on appeal the Eighth Circuit said it was erroneous. It noted that the business record rules exist because business records are likely accurate and reliable. Facebook may correctly record what users post, but the fact that something is posted on Facebook does not automatically make the content that is posted accurate and reliable. As the court put it, Facebook has “no oversight or particular interest in ensuring” the accuracy of the content users post. See generally United States v. Browne, 834 F.3d 403 (3d Cir. 2016) (similar reasoning).

A couple of quick asides about business records: First, if the only issue at a trial was whether certain content had in fact been posted – as opposed to the accuracy of that content – the business records approach would be viable. Second, there are a number of cases from courts across the country allowing surveillance videos to be authenticated as business records, a development that I view as noteworthy and about which I may write in the future.

The ruling in Farrad. Returning to Farrad, given that the government could not establish the traditional foundation for admitting photographs, and that its reliance on the business records provision failed, was the defendant’s contention that the photographs were not admissible vindicated? It was not. The Eighth Circuit ruled that the photographs were sufficiently authenticated by circumstantial evidence, including (1) that they were posted to an account that appeared to belong to Farrad, given the name on the account and other details; (2) “the photos appeared to show Farrad, his tattoos, and . . . distinctive features of Farrad’s apartment, as confirmed by police investigation”; and (3) the lack of any evidence suggesting that the photographs were inaccurate or altered. In summary, “while there were still questions about the photos that merited probing, those questions were not so glaring as to prevent the photos from clearing the relatively lower hurdle of authentication.”

Other decisions. Other courts have admitted photographs under similar circumstances and rationales. See, e.g., United States v. Vazquez-Soto, 939 F.3d 365 (1st Cir. 2019) (a postal carrier was charged with making a bogus disability claim; at trial, the government introduced photographs it found on the defendant’s ex-wife’s Facebook page that appeared to show him motorcycling, dancing, and engaging in watersports; the photographs were properly admitted because an investigator testified that he recognized the defendant in the pictures; the court stated that “[i]n determining whether the photographs were authentic, the jurors could examine the photographs and rely on their own observations of Vazquez-Soto in the courtroom”); United States v. Martin, 822 Fed. Appx. 521 (9th Cir. 2020) (unpublished) (holding briefly that “[t]he district court did not abuse its discretion in [admitting] pictures found on . . . social media accounts associated with [the defendants’] names or birthdates,” in part because “[t]he testimony about the origins of the pictures and accounts . . . and the contents of the photographs themselves” were evidence of authenticity).

Comment. My shorthand mental summary of Farrad is that the photographs were deemed authentic because Farrad posted them and they weren’t obviously fake. Reasonable minds could disagree about whether that’s the right approach. On the one hand, everyone knows that content posted on social media – including photographic content – is not always the truth, the whole truth, and nothing but the truth. Photographs might be a little bit inaccurate due to the use of a filter or the editing out of a passer-by. Or they might be wildly inaccurate, due to extensive Photoshopping or AI generation. On the other hand, authentication is a low hurdle, and it is likely true that most pictures on social media have at least some basis in reality. Furthermore, when a person chooses to post an incriminating picture of himself or herself, there may be some common sense in the idea that if you posted it, you’ve got to live with it.

As a lagniappe, and because it seems that everyone’s talking about AI all the time now, I’ll leave you with the result I got from an AI image generator when I asked for a picture of a law professor riding a dragon. Make of it what you will: