Fargo, West Fargo & Cass CountyCriminal defense informationCall 701-929-4886
Fargo and North Dakota criminal defense guide

Authenticating Social-Media Evidence in North Dakota

A screenshot can show that words or an image appeared on a screen. It does not necessarily prove which account produced the content, who controlled the…

ElementsPenaltiesEvidenceNext steps

Facebook, Instagram, Snapchat, TikTok, X, direct messages, screenshots, platform records, authorship, metadata, deleted content, deepfakes, hearsay, completeness, and trial objections

Authenticating Social-Media Evidence in North Dakota

A screenshot can show that words or an image appeared on a screen. It does not necessarily prove which account produced the content, who controlled the account at that moment, whether the exhibit is complete, whether the displayed time is correct, or whether a human being created or altered the material. North Dakota authentication law asks what the proponent claims the exhibit is and whether the evidence would permit a reasonable juror to find that claim true. Hearsay, relevance, completeness, unfair prejudice, and the original-evidence rules remain separate questions.

“It came from the defendant’s profile” is a conclusion, not a foundation. A proper analysis identifies the source device or platform return, acquisition method, account identifiers, complete conversation, metadata, edits, time-zone handling, export history, and independent facts connecting the disputed content to its alleged author.

North Dakota treats authentication as conditional relevance

N.D.R.Ev. 901(a) states the governing requirement in a single sentence: “To satisfy the requirement of authenticating or identifying an item of evidence, the proponent must produce evidence sufficient to support a finding that the item is what the proponent claims it is.” The court decides whether enough evidence exists for a reasonable factfinder to make that finding. If the threshold is met, the jury ordinarily decides ultimate authenticity and weight.

The proponent must define the claim before proving it

“This is social-media evidence” is too vague. Is the exhibit claimed to be an accurate screenshot of a public post on a certain date, an unaltered platform export, a conversation between identified users, a record generated automatically by a provider, or a statement authored by the accused? Each claim demands different support. A witness may know what appeared on a screen without knowing who controlled the account. A platform custodian may explain a stored record without knowing who held the phone.

Rule 901 gives examples, not an exclusive checklist

Rule 901(b)(1) recognizes testimony from a witness with knowledge. Rule 901(b)(4) permits authentication through appearance, contents, substance, internal patterns, distinctive characteristics, and all the circumstances. Rule 901(b)(9) addresses evidence describing a process or system and showing it produces an accurate result. Social-media proof may use one method or several together: eyewitness testimony, a forensic extraction, matching account data, private knowledge in the message, a reply understood by the recipient, platform records, or evidence of contemporaneous device possession.

Authentication is not a finding that the evidence is true

The explanatory note to Rule 901 emphasizes that authentication does not make an exhibit admissible for every purpose; authenticated material may still be hearsay. It also may be irrelevant, incomplete, unfairly prejudicial, privileged, or offered without the original-evidence foundation. A judge can find that a jury reasonably could regard a message as genuine while leaving the jury free to reject its factual assertions or the alleged author’s identity.

State v. Burton restates the threshold

In State v. Burton, 2025 ND 83, ¶¶ 20–21, the North Dakota Supreme Court explained that authentication is conditional relevance under Rule 104(b) and may be established in multiple ways. The opinion quotes the familiar rule that the proponent does not have to conclusively prove authenticity; the proof must permit a reasonable juror to find the evidence is what it purports to be. That standard is practical, but it is not meaningless. The proponent still must connect the offered item to the specific proposition for which it is presented.

State v. Henderson shows why foundation matters

State v. Henderson, 2024 ND 42, ¶¶ 9–27, reversed convictions where important financial documents, images, receipts, and video lacked adequate authentication or business-record support. An investigator’s receipt of material does not prove how its source created or maintained it. The same point applies to social-media evidence: an officer can describe downloading, receiving, or printing a file, but that testimony does not automatically establish the platform’s process, the completeness of the return, or user authorship.

A social-media exhibit contains several separate layers of proof

First: is the exhibit an accurate representation?

A witness who took a screenshot can testify when, where, and how it was captured and whether it fairly reflects what the witness saw. That may authenticate the screenshot as a depiction. It does not establish that omitted material never existed, that the application did not display a nickname assigned by the recipient, or that the visible profile belonged to the alleged author. Screen recording, original files, and contemporaneous notes may make the depiction foundation stronger.

Second: is the account correctly identified?

Display names are not unique identifiers. Handles change. Profile photographs can be copied. A page may be public, parody, shared, compromised, automated, or impersonating someone. Useful account evidence includes provider-assigned user numbers, creation dates, verified email addresses, registered telephone numbers, login records, linked devices, recovery events, name-change history, payment information, and communications proving long-term control. Even those records may identify an account subscriber without identifying the person who composed a particular message.

Third: who authored the disputed content?

Authorship can be shown circumstantially. Details known only to participants, recognizable phrasing, references to recent events, photographs not publicly available, real-time coordination, subsequent conduct, and responses that make sense in context can support attribution. Conversely, a shared password, accessible tablet, stolen phone, account takeover, scheduled post, social-media manager, or third-party application may weaken it. The question is not whether the account is generally associated with someone; it is whether the evidence supports authorship at the relevant time.

Fourth: is the exhibit complete and chronologically accurate?

A screenshot may exclude the preceding prompt, later correction, reaction, attachment, quoted reply, disappearing message, or part of a longer thread. Applications display local time, server time, relative time, or labels such as “yesterday.” Exports may use UTC. Daylight-saving changes and device settings can create apparent discrepancies. Counsel should demand the raw timestamp, time-zone documentation, thread identifier, message identifier, edit history, and records showing whether content was deleted or unsent.

Fifth: what proposition is each statement offered to prove?

A message can be offered as an alleged admission, a threat, notice, effect on the recipient, a verbal act, context for another statement, or proof that an asserted fact was true. Those are not interchangeable. The judge should identify the actual purpose and, where appropriate, give a limiting instruction. Calling every message “context” cannot become a way to place otherwise inadmissible accusations before the jury.

Screenshots, phone extractions, and platform returns prove different things

A witness screenshot preserves appearance but often loses metadata

Screenshots are useful because social-media content can change quickly. They also flatten an interactive record into an image. A screenshot may omit the full URL, internal account number, message ID, attachment file, reactions, edit marker, reply relationship, or source metadata. Cropping and compression make later evaluation harder. Counsel should seek the original image file rather than a copy pasted into a report, presentation, or word-processing document, because the original may contain capture time, dimensions, and device information.

A device extraction can preserve local application artifacts

A forensic extraction may recover application databases, cached media, notifications, tokens, account identifiers, message records, deleted remnants, and operating-system timestamps. But “the phone extraction” is not self-explanatory. Acquisition type, tool version, supported application version, lock state, encryption, parsing rules, validation, and manual interpretation matter. A generated report is an examiner’s rendering of underlying data, not necessarily the data itself. Defense review should include the forensic image or lawful working copy, hash values, extraction logs, parser warnings, and native database tables.

A provider return reflects what the platform retained and produced

Platform records may contain subscriber data, login history, stored content, media, device sessions, account changes, and provider certifications. The scope varies by service, legal process, retention, encryption, user settings, and date. A return may omit end-to-end encrypted messages, expired content, data outside the request, or information the provider no longer stores. The production’s cover letter, field guide, warrant, request parameters, and completeness limitations belong with the data.

Self-authentication can establish process integrity without proving authorship

N.D.R.Ev. 902(13) permits certified records generated by an electronic process or system, and Rule 902(14) addresses certified data copied from an electronic device, storage medium, or file when authenticated by a digital-identification process. The proponent must give reasonable written notice and make the record and certification available for inspection. A valid certification can establish that a copy matches source data or that a system generated a record. It does not prove that the accused, rather than another user, wrote a post.

Hash values test sameness, not truth

A cryptographic hash is a digital fingerprint used to compare files or datasets. Matching hashes strongly support that a copy has not changed since the reference hash was calculated. They do not prove the content was accurate when created, that an account was not compromised before collection, or that a human author is correctly identified. The record should state what was hashed, when, with which algorithm, by whom, and whether later reports or exports have their own hashes.

Chain of custody should identify every material transformation

Social-media evidence can move from a user’s screen to a screenshot, text message to police, evidence upload, report, prosecutor file, cropped trial exhibit, and courtroom display. Provider data may be unzipped, converted, indexed, filtered, decoded, translated, and exported. A defensible chain documents the original source, each transfer, storage controls, hash checks, software used, and all changes. Not every gap demands exclusion, but unexplained transformations can undermine both admissibility and weight.

Account ownership and message authorship are not the same issue

A familiar name and photograph are weak standing alone

Anyone can type a name, copy an image, or create a similar handle. The platform may display the recipient’s locally assigned contact name rather than the sender’s chosen identity. A screenshot should be examined for the exact handle, numeric identifier, verification indicator, URL, and application interface. Investigators should record how they navigated to the account and preserve the surrounding profile instead of relying on a single cropped image.

Possession of a logged-in device can support but not end the inquiry

A phone seized from a person while logged into an account is meaningful circumstantial evidence. The inference depends on when the relevant message was sent, whether others used the device, whether multiple devices were active, and whether the service permits remote posting. Lock-screen configuration, biometric enrollment, saved passwords, session logs, household access, workplace practices, and contemporaneous location may strengthen or weaken attribution.

Distinctive knowledge must truly be distinctive

A message referencing a nickname, relationship, or event may appear personal, yet the fact could be public, shared in a group, visible in earlier messages, or known by several people. Counsel should identify the supposed secret and test who actually knew it. The force of circumstantial authentication comes from the combined circumstances, not from declaring common information unique.

Conversation evidence works in both directions

Replies can show that the recipient understood who was speaking, and a sequence can contain natural continuity. But a recipient’s assumption is not always correct. Sudden changes in language, tone, knowledge, posting time, or requested payment may indicate takeover or impersonation. A complete thread may reveal uncertainty—“Who is this?”—that a selected prosecution screenshot omits.

Login data requires careful interpretation

An IP address generally identifies an internet connection, not a human. Mobile carrier translation, shared Wi-Fi, business networks, VPNs, privacy relays, and dynamic allocation complicate attribution. Device-model labels can be generic or user supplied. A successful login may reflect a saved session, background refresh, or automated service rather than active message composition. Any location or identity opinion should account for the platform’s documentation and the provider records used to interpret the log.

Proof of account compromise must be investigated, not assumed away

Warning emails, password resets, unfamiliar sessions, recovery-address changes, multifactor prompts, malware, phishing messages, reports to the provider, and contemporaneous statements may support an account-takeover issue. The absence of a formal report does not conclusively disprove compromise, particularly when a user did not know about the disputed content. Conversely, a bare claim of hacking is not self-proving. Both sides should test it against retained logs and device artifacts.

Authentication, hearsay, and completeness require separate rulings

A party statement still must be attributed to the party

N.D.R.Ev. 801(d)(2) excludes certain opposing-party statements from hearsay. That rule does not answer whether the accused made the statement. Before treating a direct message or post as a party statement, the State needs evidence from which the jury reasonably could find the accused authored, adopted, or authorized it. An account label cannot do all the work.

Incoming messages may present their own hearsay problem

A conversation often includes statements by friends, witnesses, anonymous users, bots, and reposted sources. Each statement must be analyzed. An incoming accusation is not automatically admissible for its truth merely because the accused responded. It might be admissible only to explain the response, and the court should consider redaction or a limiting instruction. If one message quotes another, N.D.R.Ev. 805 requires a justification for each layer of hearsay.

Provider records can mix machine data and user assertions

Login time or a system-assigned account number may be automatically generated. A profile name, birthday, caption, or location tag may be supplied by a user. A business-record certification can support the provider’s routine recordkeeping under N.D.R.Ev. 803(6), but it does not transform every user-supplied assertion into a business fact. Investigative notes embedded in a report also need an independent basis.

Rule 106 addresses misleading excerpts

N.D.R.Ev. 106 permits an adverse party to require introduction, at that time, of another part of a statement that fairness requires to be considered with the offered part. For social-media evidence, omitted prompts, replies, images, reaction symbols, timestamps, and corrections can change meaning. Completeness is not a license to admit an entire account; the requested material should correct the misleading impression created by the selected excerpt.

Original-evidence rules accommodate electronic information

Under N.D.R.Ev. 1001, an “original” of electronically stored information includes a printout or other output readable by sight if it accurately reflects the information. N.D.R.Ev. 1003 generally permits a duplicate unless a genuine question is raised about the original’s authenticity or admission would be unfair. Thus, “it is only a screenshot” is not automatically decisive. The real issues are whether it accurately reflects the source and whether cropping, edits, missing metadata, or a genuine authenticity dispute makes the offered copy unfair.

Summaries and illustrative aids have different rules

A party may use N.D.R.Ev. 1006 to prove the content of voluminous admissible material if the originals or duplicates are made available as required. Effective March 1, 2026, N.D.R.Ev. 107 separately governs illustrative aids used to help the factfinder understand evidence or argument. A message timeline, follower graph, or account-activity chart should be labeled and treated according to what it actually is; an advocate-created illustration is not transformed into evidence by polished graphics.

Edits, impersonation, synthetic media, and disappearing content demand source-level proof

Simple editing can create a convincing false screenshot

A fabricated social-media image does not require sophisticated software. Text, avatars, timestamps, reaction counts, and interface elements can be altered or recreated. Visual plausibility is therefore one fact, not a guarantee. Source-device data, the other participant’s device, platform records, message identifiers, media files, notification history, and consistent timestamps offer stronger corroboration than appearance alone.

Platforms permit legitimate edits that can change meaning

Some services allow captions or messages to be edited, deleted, or unsent. Others keep public edit histories; some retain internal versions not shown to users. A later screenshot may not depict what a recipient originally saw. Investigators should document the collection date, edit indicators, version history, and platform behavior applicable on that date. Counsel should not assume today’s interface and retention rules were the same when the event occurred.

Disappearing content shifts importance to contemporaneous collection

Ephemeral messages may be captured by a recipient, cached in notifications, preserved in a backup, recovered from local storage, or retained briefly by a provider. Each source has limitations. A recipient’s photograph of a screen may accurately preserve visible content even when no provider copy exists, but the foundation should explain the device, application, sender identifier, timing, and surrounding exchange. The inability to obtain a platform record affects corroboration; it does not create a categorical rule for or against admissibility.

Deepfakes require a precise claim about what was synthesized

“AI generated” can refer to a wholly synthetic image, face replacement, cloned voice, altered background, automated caption, enhanced resolution, or ordinary filter. Counsel should identify the suspected manipulation instead of using one label. The original media file, encoding history, generation metadata, prior versions, upload source, account logs, camera artifacts, and independent witnesses may matter. Recompression by a platform can strip metadata and complicate analysis.

Expert opinions must fit the actual forensic question

N.D.R.Ev. 702 requires expert testimony to rest on sufficient facts and reliable methods reliably applied. An examiner who verifies a file hash may not be qualified to identify synthetic imagery; a visual-effects specialist may not be able to attribute an account. The proponent should disclose tools, versions, validation studies, error limitations, comparison sources, and all processing. A classifier score should not be presented as certainty without explaining what the model measures and how it performs on comparable material.

Enhancement must not be confused with recovery

Upscaling, sharpening, denoising, frame interpolation, color adjustment, and audio cleanup can make media easier to view. Generative enhancement may add predicted detail that never existed in the source. The jury should know which operations were used, whether pixels or sounds were synthesized, and whether the original remains available. Side-by-side presentation and a reproducible processing log can prevent an enhanced exhibit from being mistaken for untouched source evidence.

Trial preparation should match each exhibit to a witness and purpose

Build an exhibit-specific foundation chart

For every proposed post, message, image, video, or account record, identify the exhibit number, exact evidentiary claim, source, collector, original location, account identifier, relevant date and time zone, hash, transformations, authenticating witness or certification, hearsay basis, completeness material, and requested limitation. A chart exposes gaps before trial and prevents a single generic “social media” ruling from obscuring different problems.

A recipient can authenticate receipt without proving platform mechanics

A participant may testify that a message appeared in a conversation, that the exhibit accurately depicts it, and that surrounding circumstances identified the sender. That testimony can be powerful. It does not make the witness a custodian for the provider, an expert on server logs, or a forensic examiner. The foundation should remain within the witness’s personal knowledge under N.D.R.Ev. 602.

An officer’s collection testimony has defined limits

An officer can explain viewing a public profile, receiving a screenshot, serving process, downloading a return, or seizing a device. The officer ordinarily cannot establish provider business practices merely by possessing the resulting file. Henderson is important here: custody after receipt is not a substitute for source foundation. If the State relies on a provider certificate or forensic examiner, the exhibit and notice should be disclosed early enough for meaningful review and objection.

Objections should identify the missing link

“Foundation” alone may be too general to obtain a focused ruling. Counsel should state whether the problem is inaccurate depiction, account identification, authorship, incomplete thread, unexplained timestamp, missing source file, unreliable process, hearsay, embedded hearsay, best evidence, unfair prejudice, expert qualification, or chain of custody. N.D.R.Ev. 103 requires a timely objection or motion to strike stating the specific ground unless apparent from context.

Rule 403 can limit misleading presentation

Under N.D.R.Ev. 403, relevant evidence may be excluded when unfair prejudice, confusion, misleading the jury, delay, or needless cumulation substantially outweighs probative value. Inflammatory posts, large follower counts, unrelated political speech, sexual material, music lyrics, or offensive memes may invite character reasoning unrelated to the charged act. Redaction, a limiting instruction, a neutral account label, or a smaller set of representative exhibits may preserve legitimate proof without turning trial into a referendum on online identity.

Demonstratives should not exaggerate certainty

A prosecutor-created chat bubble, reconstructed profile, animated timeline, or network diagram can look more authoritative than the underlying exhibits. Dates, colors, avatars, and arrows may contain assumptions. Rule 107 now supplies a framework for illustrative aids, while Rule 1006 controls evidentiary summaries of voluminous records. Counsel should insist that the court and jury know which category applies and that disputed authorship is not visually presented as an established fact.

Defense counsel should preserve, collect, and test social-media evidence early

Send targeted preservation demands without altering the evidence

Content can disappear through user action, routine retention, expiring messages, account closure, or device replacement. Counsel should promptly identify relevant accounts, devices, date ranges, conversations, media, login history, and recovery records. A preservation request does not itself compel production, and defense access to provider content can raise Stored Communications Act issues. Lawful collection should be coordinated with subpoenas, court process, consent, discovery, and the client’s obligations. The client should not delete posts, contact witnesses about changing content, or log into another person’s account.

Collect both sides of the conversation when lawfully available

The accuser’s screenshot may omit messages retained on the client’s device, and the client’s copy may omit expired or deleted items visible elsewhere. Compare sender and recipient devices, provider returns, cloud backups, notification databases, exported archives, and witness captures. Differences may result from syncing, deletion, editing, application version, or manipulation. They should be documented rather than silently reconciled.

Demand native data and collection documentation

Discovery should include original screenshot files, uncropped versions, screen recordings, forensic images or reviewable working copies, extraction reports, audit logs, hash manifests, parser warnings, platform productions, certifications, legal process, field definitions, time-zone information, account history, investigator notes, demonstrative source files, and every transformed exhibit. A PDF printout is often insufficient to evaluate metadata, message relationships, or deleted records.

Reconstruct the account and device timeline

Create a chronology of account creation, username changes, password resets, login sessions, device acquisitions, phone-number changes, recovery events, disputed posts, edits, deletions, seizures, extractions, and provider production. Add known travel, work, witnesses, surveillance, and device-location records. Attribution often turns on whether control evidence aligns with the exact minute of the disputed act rather than with general account ownership.

Use an expert only when the disputed issue requires one

Some cases need no expert because a participant can explain an ordinary exchange. Others require forensic examination of databases, deleted artifacts, manipulated media, login records, time conversions, or synthetic-content claims. Define the assignment carefully. The expert should receive source data and comparison material, document methods, preserve reproducibility, state limitations, and avoid offering a legal conclusion or an identity opinion beyond the technical evidence.

File focused motions in limine and preserve trial objections

A pretrial motion can request exclusion, a foundation hearing, complete production, a Rule 106 completion, redaction, expert disclosure, or limits on demonstratives. The motion should address each exhibit category and explain the precise defect. If the ruling is preliminary, renew the objection when the exhibit is offered. If the court admits evidence for a limited purpose, request an instruction tailored to that purpose and object if argument exceeds it.

Do not let admissibility replace reasonable doubt

Meeting Rule 901’s threshold does not prove guilt. The jury may still find that an account was shared, a message was ambiguous, a meme was reposted without endorsement, a statement was sarcasm, a timestamp was misunderstood, or an exhibit was incomplete. The defense theory should connect evidentiary limitations to the elements: identity, knowledge, intent, agreement, threat, possession, or participation. Technical cross-examination is most useful when it explains why a claimed inference is not reliable enough to satisfy the State’s burden.

Frequently asked questions about social-media evidence in North Dakota

Is a screenshot automatically admissible?

No. The proponent must authenticate what the screenshot is claimed to show, and hearsay, completeness, relevance, original-evidence, and Rule 403 objections may remain.

Does a username prove who wrote a message?

No. A username is one circumstance. Account identifiers, control history, device evidence, distinctive content, platform logs, and corroborating conduct may be necessary to support authorship.

Must Facebook, Snapchat, Instagram, or TikTok send a witness to court?

Not always. Testimony from a participant or other knowledgeable witness may authenticate some evidence, and qualifying provider or electronic records may be certified under Rules 902 and 803. A certification does not prove user authorship.

Can a recipient authenticate a direct-message conversation?

Potentially. The recipient can describe personal participation, how the messages appeared, surrounding circumstances, and why the sender was recognized. The adequacy of that foundation depends on the claim and the facts.

Are a defendant’s social-media statements hearsay?

Qualifying opposing-party statements are excluded from hearsay under Rule 801(d)(2), but the State must still provide sufficient evidence that the defendant made, adopted, or authorized the statement.

What if the screenshot leaves out part of the conversation?

Rule 106 may require another part to be introduced when fairness requires it to be considered with the excerpt. Omitted context also bears on authentication, meaning, and weight.

Does the best-evidence rule prohibit screenshots?

Not categorically. Electronic originals include readable outputs that accurately reflect the information, and duplicates are generally admissible. A genuine authenticity question, unfairness, or inaccurate and incomplete reproduction can change the analysis.

What is metadata?

Metadata is information about a file, message, or record, such as identifiers, timestamps, dimensions, encoding, source application, or creation and modification details. Its availability and meaning depend on the system.

What does a hash value prove?

A matching cryptographic hash supports that the compared digital items are identical. It does not prove who created the content, whether it was true, or whether it had been manipulated before hashing.

Can deleted or unsent messages be used?

Sometimes. They may survive in device artifacts, notifications, backups, recipient copies, or provider data. The proponent must authenticate the recovered material and explain the source and recovery method.

How can the defense investigate an account-takeover claim?

Review login sessions, device history, security alerts, password and recovery changes, multifactor prompts, provider reports, malware evidence, and conduct surrounding the disputed message.

Does an IP address identify the author?

Usually not by itself. It may identify a connection or subscriber at a time, but shared networks, carrier systems, VPNs, dynamic assignment, privacy services, and automated sessions complicate attribution.

Can AI-generated images or audio be admitted?

Potentially, depending on relevance and purpose, but disputed synthetic or altered media requires a source-specific foundation. Expert reliability, original files, processing history, metadata, and corroboration may be critical.

Can police authenticate evidence simply by saying they received it?

Receipt can establish part of custody. It does not necessarily establish how a platform created the record, whether the return is complete, or who authored user content.

Are platform records automatically business records?

No. The proponent must satisfy the applicable rule or certification requirements. Even when provider-generated fields qualify, user-supplied assertions and embedded third-party statements require separate analysis.

When is a social-media expert needed?

An expert may be useful for forensic extractions, databases, deleted artifacts, manipulation, synthetic media, login interpretation, or complex time conversion. Ordinary participant testimony may be enough for a straightforward exchange.

Can inflammatory posts be excluded even if relevant?

Yes. Rule 403 permits exclusion when specified dangers, including unfair prejudice or misleading the jury, substantially outweigh probative value. Redaction or a limiting instruction may be alternatives.

What should defense counsel request in discovery?

Native files, uncropped captures, complete threads, device extractions, hashes, logs, certifications, platform returns, legal process, field guides, edit and login history, collection notes, time-zone details, and transformed exhibits.

Official authorities cited

Legal information, not legal advice. Social-media admissibility depends on the precise evidentiary claim, source, collection, platform, account identifiers, device control, surrounding communications, metadata, completeness, edits, provider practices, certifications, hearsay purpose, expert method, trial objections, and current law.


Call nowChoose a time