AI Social Media Surveillance in Michigan Injury Claims (2026)

Knowledge Base · AI & The Law

AI Social Media Surveillance in Michigan Injury Claims (2026)

By Attorney Manny Chahal · Updated June 2026 · Reading time: ~9 min · Educational overview — not legal advice

If you have filed a Michigan no-fault or third-party injury claim in 2026, an adjuster may well have fed your name into an AI-driven social media monitoring tool. This guide explains what those systems collect, how Michigan courts treat the resulting evidence under MRE 901 and Michigan personal injury case law, and where claimants most often hurt their own cases.

Why Insurers Are Investing in AI Surveillance

Open-source intelligence has moved from a defense-counsel side project to a routine adjuster workflow. Industry reporting and trade publications indicate that many of property and casualty carriers now run some form of automated social-media monitoring on injury claims at intake, with periodic refreshes triggered by claim value, treatment milestones, or attorney representation. The tools combine name and email matching, facial recognition across public photographs, geolocation extracted from posts, and natural-language scoring of captions and comments. The output is a dossier the adjuster reviews alongside medical records.

The economic logic is straightforward. An insurer that spends modest cost on automated scraping can sometimes reduce a five- or six-figure payout by reframing the claimant’s reported limitations. Even when the content is innocuous, the file becomes a tool to anchor settlement negotiations and to seed deposition questions.

Practical reality: Adjusters typically begin gathering social-media content within hours of a first-notice-of-loss call. Setting an account to private after the claim is filed does not erase what was already collected, and it may itself become a discovery issue.

What These Systems Actually Pull

Modern claim-investigation platforms reach well past Facebook profile photos. A 2026 dossier on a Michigan PIP claimant may include:

  • Public posts and photos across Facebook, Instagram, X, TikTok, YouTube, LinkedIn, Pinterest, and Reddit, including content where the claimant is tagged by friends or family.
  • Geotagged check-ins that suggest travel, gym attendance, or recreational activity inconsistent with reported limitations.
  • Comment trails on third-party posts (e.g., a high-school reunion thread where the claimant mentions a recent trip).
  • Wearable and app data referenced in posts — step counts, Strava routes, Apple Watch summaries, Fitbit milestones.
  • Public reviews the claimant has written for restaurants, gyms, or hotels, often with dates and locations attached.
  • Marketplace activity — a Facebook Marketplace listing for a road bike or kayak posted three weeks after a reported low-back injury draws scrutiny.

The tools also flag content from connected accounts: a spouse’s Instagram showing a family vacation, a child’s TikTok with the claimant lifting a cooler, a friend’s Facebook tag at a wedding reception. The AI is doing two things at once — it is gathering broadly, and it is scoring relevance against the specific injuries pled in the claim or complaint.

How Michigan Courts Treat the Evidence

Authentication Under MRE 901

Before any social-media exhibit reaches the jury, the proponent must satisfy MRE 901(a) by producing evidence sufficient to support a finding that the item is what the proponent claims it to be. The Michigan Rules of Evidence (current PDF reflects orders effective January 28, 2026) do not single out social media. The same authentication menu in MRE 901(b) applies — testimony from a witness with knowledge, distinctive characteristics, comparison by an expert, or process and system evidence under MRE 901(b)(9).

In practice, defense counsel typically authenticates a Facebook post through one of three routes: (1) the claimant’s own deposition admission that the account is theirs, (2) a subpoena to the platform under the Stored Communications Act with a certification of business records, or (3) testimony from the OSINT vendor or paralegal who captured the page along with a Page Vault or similar hash record.

Authentication Is Only the First Hurdle

Authentication is not the same as admissibility. Once a post is shown to be what it appears to be, a Michigan trial court still asks whether it survives hearsay rules, MRE 403 prejudice analysis, and (where the post is being offered to prove a state of physical capability) MRE 702 reliability for any expert opinion drawn from it. A still photograph of a claimant smiling at a barbecue, standing alone, says very little about whether MCL 500.3135 serious-impairment standards are met. The defense usually has to pair such content with medical-records contradictions or surveillance video to make the inference stick.

Discovery Scope Under MCR 2.302

Michigan’s discovery rules expressly reach electronically stored information. MCR 2.302(B)(1) permits discovery of any non-privileged matter that is relevant to a party’s claim or defense and proportional to the needs of the case. After the 2020 discovery-rule overhaul, MCR 2.302(A) requires initial disclosures that include the categories of ESI a party may use to support its claims or defenses. Private social-media content is not automatically off-limits — a Michigan court can compel production where the defense makes a threshold showing of relevance, and the proportionality factors do not preclude review. Blanket “all my Facebook content for the last five years” requests are usually narrowed by the court to the period and topics genuinely relevant to the injuries pled.

The Spoliation Trap

One of the most damaging unforced errors a Michigan injury claimant can make is deleting posts after litigation is reasonably anticipated. Michigan’s spoliation doctrine traces back to Brenner v. Kolk, 226 Mich App 149 (1997), where the Court of Appeals held that a party has a duty to preserve evidence even before suit is filed, so long as the party knows or reasonably should know the evidence may be relevant to anticipated litigation. The duty has been applied to surveillance video, vehicle parts, business records, and — in published opinions across other jurisdictions and an emerging body of Michigan federal decisions — to social-media content.

When a Michigan trial court finds that material evidence was spoiled, Brenner directs it to fashion a sanction that denies the spoliating party the fruits of its misconduct without going further than necessary. Common sanctions include an adverse-inference jury instruction, exclusion of related testimony, monetary sanctions, or in egregious cases dismissal of the affected claim. A claimant who deletes the cycling-vacation photos after filing an MCL 500.3105 PIP demand can find that the adverse-inference instruction does more damage than the deleted content ever would have.

The safer path: Once a claim is filed or reasonably anticipated, lock down privacy settings going forward, but do not delete, archive, or “scrub” historical content. Document the privacy-setting change with a dated screenshot.

How the AI Layer Changes the Old Playbook

IssuePre-AI Era (through ~2018)AI-Enabled Era (2024–2026)
Trigger for reviewAdjuster suspicion or large claim valueAutomatic at intake on most claims; periodic refresh
ScopePublic Facebook profile, maybe LinkedInCross-platform graph including tagged photos, family, geolocation, wearable data
Identity matchingName + city; manual verificationFacial recognition across public photos; email/phone reverse lookup
OutputLoose printout in the fileScored dossier flagging “high-value” inconsistencies for adjuster review
Defense at trialAuthenticate via deposition admissionSame plus vendor-witness foundation, hash logs, and Page Vault certifications
Plaintiff riskEmbarrassing photo at depositionPattern arguments built from dozens of micro-data-points

Deepfakes, Manipulation, and the 2025 Michigan Statute

The other side of the AI surveillance question is what to do when the content itself is fabricated. Michigan enacted the Intimate Deep Fakes Act in 2025 (PA 11 of 2025), defining a “deepfake” as audio, video, or digital content that has been substantially manipulated or generated by artificial intelligence to depict an identifiable individual in a way the individual did not actually engage in. The Act primarily targets non-consensual intimate imagery, but its definitional language is influencing how Michigan courts think about generative-AI content more broadly. When defense counsel offers a surveillance still that looks unusually clean or shows the claimant in an oddly posed activity, claimant’s counsel may now layer a deepfake-detection objection on top of the MRE 901 authentication challenge. Expert testimony from a forensic image analyst is usually required to support such a challenge.

What Michigan Injury Claimants Should Do Right Now

The following list is general guidance and is not a substitute for advice from counsel familiar with the specific claim:

  • Lock down privacy settings going forward. Set Facebook, Instagram, TikTok, and similar profiles to private. Do not accept new friend requests from people you do not know — “sock-puppet” accounts created by claim investigators are widespread.
  • Do not delete historical content. The duty to preserve under Brenner v. Kolk attaches before the lawsuit is filed.
  • Pause posts about activity, travel, and physical milestones while the claim is pending. A weekend road-trip photo, even if it post-dates the recovery period, can be re-cast at deposition to undermine PIP wage-loss continuity under MCL 500.3107(1)(b).
  • Tell your friends and family not to tag you in photos or check you in to locations.
  • Capture screenshots of your own profiles on the day of the claim. A claimant-side preservation record protects against later spoliation arguments and also documents what the insurer could have seen.
  • Tell your attorney about every social-media account you have, including dormant ones. The defense will find them; the only question is whether your lawyer learned about them first.

Frequently Asked Questions

Can the insurance company really see my private Instagram?

Not directly, in most cases. A private account blocks general scraping, but the insurer can still see content reposted publicly by your tagged friends, family, or employer. In litigation, the carrier can serve a discovery request under MCR 2.302 for relevant private posts, and a Michigan court may order production if the request is targeted to the injuries pled and not a fishing expedition. The carrier can also subpoena the platform itself, though the Stored Communications Act restricts what platforms will produce without your consent.

Should I deactivate my Facebook account after a Michigan injury claim?

Generally no. Deactivation looks indistinguishable from deletion in a spoliation context, and Brenner v. Kolk, 226 Mich App 149 (1997), supplies the duty to preserve. The conservative move is to lock down privacy settings going forward, stop posting new content related to your activities, and preserve a screenshot record of what was on the profile as of the claim date.

My friend posted a photo of me at a wedding three months after the accident. Can the insurance company use it?

Yes, if the post is publicly viewable or surfaces in formal discovery. Authentication under MRE 901 is straightforward when you can be identified and the friend can confirm the date. Admissibility usually still depends on whether the post tends to prove a material fact — for example, whether you can perform an activity inconsistent with your reported impairment under MCL 500.3135. A single photograph rarely carries that weight; a pattern of photographs often does.

What if I think the insurer’s AI tool got the wrong person?

Identity errors are common with facial-recognition matching. Raise the issue early. Demand the vendor’s match-confidence score and the underlying source URLs in discovery, and consider an MRE 901 challenge plus an MRE 702 reliability objection if the carrier intends to use the content at trial. Forensic image-analyst testimony may be necessary in close cases.

Does any of this change my filing deadlines?

No. Social-media surveillance is a litigation-conduct issue. Your underlying filing windows are unchanged: no-fault PIP claims are bounded by the one-year-back rule under MCL 500.3145 and a one-year notice/suit framework; third-party tort claims for bodily injury generally run on the three-year statute under MCL 600.5805; and comparative-fault allocation under MCL 600.6304 still applies.

Can my own lawyer use AI to push back?

Yes, with appropriate verification under MRPC 1.1 competence and MRPC 5.3 supervision standards. Claimant-side tools can map insurer surveillance dossiers, run independent OSINT preservation, and identify forged or AI-manipulated content. Any AI-assisted work product offered as evidence should be authenticated by a competent witness and never cited without manual verification — the published sanction record on hallucinated citations is a cautionary tale.

Injured in Michigan? Don’t face this alone.

Free, confidential review with Attorney Manny Chahal. No fee unless we recover.

Call 1-844-624-2425