Summary

  • The Indiana case against Georgetown officer Justin Craig turns on the previously undefined boundary of authorized use of the shared Flock plate-recognition network.
  • A Georgetown police internal review found Craig’s hundreds of searches unrelated to any police investigation and mislabeled under categories including traffic infractions, narcotics, assault/battery, theft and stolen vehicles, per documents reported by WLKY.
  • Flock’s CEO announced in August that law enforcement agencies must switch on audit features, moving the definition of misuse into the product’s architecture.
  • Defense attorney Matthew Lemme said the alleged conduct “does not violate any law or department policy” and announced that Officer Craig will file a motion to dismiss, placing the definitional contest before a court.

The criminal case against Georgetown, Indiana, police officer Justin Craig — charged with official misconduct and two counts of fraud over Flock camera searches that documents show were directed at his wife’s license plate more than 80 times and the plate of a “previous acquaintance” of hers nearly 250 times between January and July — turns on a concept the reported record does not show was defined in advance by Indiana statute or department policy: what counts as misuse of a shared plate-recognition network. A Georgetown police internal review found the searches unrelated to any police investigation and intentionally mislabeled to appear justified; a Washington Post reporter triggered the investigation by calling the police chief with a tip, the documents show. Flock’s CEO has announced features intended to curb police abuses of the technology, and defense attorney Matthew Lemme said the alleged conduct “does not violate any law or department policy” and that Officer Craig will file a motion to dismiss. The parties have framed the question in definitional terms: whether a query into the shared network is governed by the judgment of the officer entering it, a departmental rule, a statute, or the vendor’s product design — and whichever answer prevails determines who carries the cost when a query serves no investigation.

The reported record

Justin Craig, an officer with the Georgetown police department, was arrested on charges of official misconduct and two counts of fraud, according to documents reported by WLKY and set out in The Guardian’s account. The documents show that Craig searched his wife’s license plate more than 80 times and the plate of a “previous acquaintance” of hers nearly 250 times, between January and July.

A Georgetown police internal review found that the searches Craig conducted on the Flock system were unrelated to any police investigations, and that he had intentionally mislabeled those searches to make them appear justified; the labels used on the searches included traffic infractions, narcotics, assault/battery, theft and stolen vehicles. A reporter from the Washington Post triggered the investigation after calling the Georgetown police chief with a tip, the documents show.

The two search subjects — Craig’s wife and a “previous acquaintance” of hers — are identified in the documents but not named in the article.

The contested concept

The underlying facts are individual; the question they put in play is not. Philosophers of conceptual engineering — most prominently Herman Cappelen — have argued that concepts can become defective when they fail to serve their purposes, and that revising the concept can be more tractable than repairing the world around it. The case offers a concrete instance: the concept of misuse was left undefined when the system was deployed, and the parties are now attempting to define it in different registers, toward different purposes. The case has become less a straightforward misconduct matter than a contest over where that boundary gets drawn — by the legislature, the police department, the technology vendor, or a court.

Authorization on systems of this kind is defined procedurally rather than substantively: an operator with credentials enters a plate and selects a category, and the category is supplied by the operator. Three functions the current concept does not perform follow from that design. It cannot distinguish at the moment of entry between investigative purpose and personal purpose. It produces a record of the operator’s stated justification rather than of an independently verifiable one — which is why, on the documents reported by WLKY, the labels and the review’s finding diverge: a label field recorded justifications that diverged from the investigation record. And it gives the person queried no way to learn that they were queried; the article describes no notification mechanism for the subjects of these searches.

Candidate definitions exist at three sites. At the statutory level, the case proceeds through general criminal categories — official misconduct and fraud — rather than a rule written for personal queries of plate data; the reported record identifies no statute or policy specifically directed at personal-use queries of the plate network, and the defense’s position is that the conduct alleged “does not violate any law or department policy.” At the policy level, the internal review’s finding that the searches were deliberately mislabeled presupposes a departmental rule about the accuracy of labels; what the reported record does not show is a rule against the underlying personal-use queries themselves. The labels were, on the record, the only gate between authorized and unauthorized use, and if the fraud charges rest on the falsification of that gate, as the reported sequence suggests, the legal core of the case is the corruption of the system’s record rather than the searches alone. Whether the department’s own labeling practice was itself an enabling condition is not addressed in the account. At the technical level, Flock has begun defining misuse inside the architecture: in August, the company’s CEO announced that law enforcement agencies would be required to switch on features that could help “audit the way individual officers use its tracking capabilities.”

The ameliorative purpose a revised concept would serve is specific: candidate revisions make a query answerable to a purpose that exists outside the operator — an open matter, a documented request, an audit checkable against records rather than against labels. Candidate revisions follow in ascending order of force: system-enforced linkage between a query and an open case identifier; mandatory audit of query logs (the direction of Flock’s August announcement); and statutory purpose limitation — the data-protection principle that data should be processed only for specified, explicit purposes — defining a violation. The defense statement already names the last of these as the proper venue, arguing that if the conduct should be prohibited, the answer is to write that law or policy, and that the public conversation about privacy and practice should precede charging officers.

A fourth engineering move is visible in the surrounding public debate rather than in the case: defining misuse from the perspective of the scanned. One design implication, drawn from an ameliorative purpose, is that a system built around the subject’s interest rather than the investigator’s convenience could include default-on audit logs; purpose fields that require case identifiers and reject labels the system cannot substantiate; retention limits on the location histories the network accumulates; and a defined process for subjects to learn that they were searched. None of these elements requires new technology; each is a choice about where the burden of justification sits. The ameliorative purpose such a design serves — protecting the people whose movements the network records — differs from the purpose the current design serves, which is investigative utility, a purpose Flock states explicitly. The two purposes are in tension; the case is, in part, a contest over which of them the definition of misuse serves.

A license plate is displayed in public; as background, the legal tradition treats what is voluntarily exposed to public view as largely outside the Fourth Amendment’s protection. In a line of 1970s cases known as the third-party doctrine, the Supreme Court held that information shared with a third party loses constitutional protection — in the leading case, Smith v. Maryland (1979), a dialed telephone number. In 2018, the Court declined to extend that doctrine to long-term cell-phone location records, holding in Carpenter v. United States that the government generally needs a warrant to obtain them. Courts have divided over how that logic applies to automated plate readers, which observe something public but accumulate a location history at considerable scale — about 2bn scans monthly across roughly 120,000 cameras, per the figures reported, in a network serving more than 4,800 law enforcement agencies and nearly 1,000 businesses. The Georgetown case itself proceeds under state misconduct and fraud statutes, not the Fourth Amendment, but that unsettled boundary is the conceptual ground against which the definition of misuse is being contested.

Who benefits

The classical maxim cui bono — who benefits — attributed to a Roman figure named Cassius Longinus, asks, before assessing an argument, to identify whom it serves. Every available framing of this case allocates accountability differently.

The framing under which the state is proceeding — an officer used his access for purposes an internal review found unrelated to any police investigation, and intentionally mislabeled the queries — locates fault in one individual. It keeps the vendor that designed the access, the department that administered it, and the legislature that did not write a specific rule at a distance from the cost. For the department specifically, the internal review and the arrest are on the record as corrective action, which addresses the reputational exposure of an audit that might find more, amid what the article describes as bipartisan backlash over privacy violations and data-sharing.

The defense’s statement inverts the distribution, naming Flock’s business model — the same technology sold to “cities, homeowners associations, and shopping malls” — and arguing that criticism of the technology was displaced onto individual officers. For the vendor, growth depends on renewals across a heterogeneous customer base of agencies, homeowners associations and malls, and in this framing misuse can be attributed to named individual users. Whatever the legal merits of either account, both are distributional claims about where the cost of the system’s design should fall.

At the query level the incentive structure is frictionless: a category label costs less effort than a case citation, which is the mechanism the review’s mislabeling finding describes.

The symmetry runs in both directions. A systemic diagnosis — that the architecture made hundreds of personal queries over seven months possible without triggering review — is available to the officer’s defense as an exculpatory account and to the technology’s critics as an accusatory one. That the same structural argument serves both positions indicates how underdetermined the definition is; it is not itself a judgment on either use.

The counterparty’s loss pathway runs to the two individuals whose plate records were queried hundreds of times — the article identifies no mechanism by which either would learn of it, and neither is quoted in the reported record; the investigation was not opened at their request but on a reporter’s tip. The larger class bearing the ambient cost is the driving public: the accumulated location history is treated in the case as background rather than as an element, and the network is also, per the article, “frequently used by US Immigration and Customs Enforcement (ICE),” whose tactics under the second Trump administration have drawn nationwide criticism. Structurally, anyone moving through the network operates in a system whose queries are recorded but not governed.

The map of interests

The stakeholder-salience framework developed by Mitchell, Agle and Wood classifies parties by power, legitimacy and urgency.

  • The two search subjects — Craig’s wife and a “previous acquaintance” of hers — are dependent stakeholders: they hold a legitimate interest and, on the alleged facts, an urgent one, but no power to enforce either. On the reported record they have no formal role in the criminal proceedings; their dependence on proxies is the operative feature — a journalist, a police chief and a prosecutor each take up the matter.
  • The Washington Post reporter holds informal influence not visible on any governance chart: the power to trigger the official process without formal status within it.
  • The Georgetown police department is, in this matter, definitive — power over the officer, legitimacy as the investigating body, urgency supplied by the tip and the internal review. Its structural position is that its review and the arrest demonstrate corrective action.
  • Flock holds decisive power over the system’s design and over contract terms with more than 4,800 law enforcement agencies and nearly 1,000 businesses; its legitimacy in the public-safety domain has been contested amid bipartisan criticism; the August announcement reflects the urgency of that contest. By the framework’s terms it sits at the boundary of the dominant and dangerous categories — capable of acting decisively, vulnerable to challenges to its standing — depending on whether its legitimacy claim is credited. Its loss path would be legislation at federal or state level; its publicly stated position keeps governance inside the product.
  • The Indiana legislature is the latent stakeholder: it holds power and legitimacy, but the reported record shows no urgency — no specific statute has been cited by either side, and the defense has explicitly invited the legislative fix. The legislature’s absence from the case is the load-bearing gap in it. Policymakers, facing what the article describes as bipartisan criticism, hold the option the defense itself proposes.
  • ICE is a dominant stakeholder in the network as a whole — federal power and institutional legitimacy, with urgency that varies with enforcement priorities — though it is not a party to the Georgetown case; its place in the network’s data flows is part of why the privacy debate extends beyond this prosecution. Downstream users depend on continuity of access, which any tightening of the query standard would put in question.
  • Flock’s nearly 1,000 business customers — homeowners associations and shopping malls — are adjacent-domain stakeholders: if regulation of law-enforcement use hardens through audit mandates or new statutes, the business channel would remain governed by contract rather than criminal-justice policy, a possibility the defense’s own statement anticipates by naming them.

The parameter whose alteration would shift the distribution

One parameter governs how that allocation moves: whether the label field remains officer-entered free text or becomes validated against an open matter. Alter that field and the audit finding becomes difficult to generate, the individual-charge pathway becomes harder to isolate from institutional responsibility, and the vendor’s exposure moves from user conduct to product design. Leave it as is, and every downstream use — reported cases include officers accused of using the cameras to stalk former romantic partners and to track down individuals seeking abortions, and the network’s frequent use by ICE — is governed the same way.

Adoption and coordination

A community adopting a revised standard would gain verifiable purpose records and a defensible line between investigative and personal use; it would lose query speed and some officer discretion. The adoption problem is coordination: roughly 4,800 agencies, a shared commercial network, private business customers and 50 state legislatures would have to move in concert, and no single department adopts a stricter standard unilaterally while neighbors and businesses continue to share the same feeds.

The stakes of the definition

The class of potential future subjects is defined by the prior reports the article surveys: officers in other states “accused of using the cameras to stalk former romantic partners and track down individuals seeking abortions.” Those reports mark what the definitional contest is worth — the difference between conduct prohibited in advance and conduct litigated only after the fact.

Because the concept of misuse was not defined in advance — at statute, policy or design — the parties with power over its definition are now defining it through the case itself. The state defines it through general criminal statutes applied to specific conduct. The department defines it through its internal review. The vendor defines it through technical audit defaults. The legislature is invited to define it prospectively. The defense asks a court to declare that the conduct, as alleged, was never prohibited, and has announced a motion to dismiss. Each move allocates the same cost to a different party — the officer, the vendor, the agencies, the public purse. The definition arrived after the design, after the deployment, and after the searches; the Indiana charges will be decided under the rules as they stand, which is precisely the gap each party’s public statement is aimed at.

Additional considerations

  • The legal and scholarly background referenced above — Smith v. Maryland, Carpenter v. United States, the attribution of the cui bono maxim to Cassius Longinus, and Herman Cappelen’s work on conceptual engineering — derives from the wider legal and philosophical literature; external verification of these references was not completed.
  • Jurisdiction-specific rules on victim standing in Indiana criminal proceedings were not verified.
  • The article dates Flock’s audit-feature announcement only as “In August,” without a year.

Analytical techniques used in this piece

This analysis applies the methods below. Each links to a short, plain-English explainer you can read and reuse.

Conceptual Engineering
Asks not just what a concept means but what it should mean, and re-engineers it.
Cui Bono — Who Benefits
Asks who gains and who pays from a state of affairs, decision, or claim.
Stakeholder Mapping
Charts the parties to a situation — their interests, power, and alignments.