Using In Rem Jurisdiction to Ameliorate Artificial Intelligence Difficulties

November 06, 2024      New York Law Journal

By Jonathan Bick Bick is counsel at Brach Eichler in Roseland, and chairman of the firm’s patent, intellectual property, and information technology group. He is also an adjunct professor at Rutgers Law School and Pace Law School.

Artificial Intelligence (AI) may cause harm including consumer privacy violations, physical harm via automated cars or drones and legal difficulties such as infringement.  Since an AI is not a legal entity, it may avoid being held accountable for its bad acts because it cannot be a party to a court proceeding.  However, an AI may be the object of a court proceeding—more particularly, an aggrieved party may institute in rem proceedings, which would place the AI under the jurisdiction of a court and that court may transfer the AI to the plaintiff and thereby ameliorating the difficulties associated with an AI's bad acts.

Contrasting in personam jurisdiction and in rem jurisdiction, the former allows a court to obtain power over the defendant personally and unlimited recovery, while the latter allows a court to obtain power over specific property and limited recovery.  In short, in rem jurisdiction allows a lawsuit against the property itself, and thereby allows a court secure jurisdiction over property without a legal entity as a party to the jurisdiction.

An in rem legal proceeding adjudicates the rights to a specific piece of property for every potential rights holder, even potential rights holders who are not named in the lawsuit.  Generally, an in rem proceeding must be commenced in the jurisdiction where the subject property is located.  The presence of the subject property in the forum state generally fulfills any due process concerns for binding out-of-state claimants to the court's judgment.  Nevertheless, notice of litigation still must be given to all entities known to have interests in the res.

The concepts of in rem (and quasi in rem jurisdiction) incorporates the court's power to adjudicate the rights to a given item of property, including seizing it, holding it and transferring it. Most courts accept the use of in rem and quasi in rem jurisdictional bases where the property is the subject matter of the litigation or at least a significant ingredient in it (see for example, Spungin v. Chinetti Int'l Motors, 515 F. Supp. 31 (1981)).  It may be argued that if the AI software was the cause of the damage at issue, then the application of either in rem and quasi in rem jurisdictional is justified.

AI is software.  AI software differs from traditional software because a computer rather than a programmer writes the algorithm (the element of the software which tell the computer what to do). Software is personal property, though it may be tangible or intangible personal property.  While in rem jurisdiction has traditionally been associated with tangible property courts have recognized the application of in rem jurisdiction to intangible property as well (see for example, Rental Op. Co. v. V.I. Water & Power Auth., 515 F. Supp. 3d 1237 (2021)).

Consequently, for in rem jurisdictions purpose, it does not matter if the AI software is tangible or intangible personal property.  However, the exercise of in rem jurisdiction over intangible property may cause an additional level of complexity as to where the res (the AI software) is located.

In any case, the location of the res (AI software at issue), only determines which jurisdiction's in rem laws apply, and doesn't determine if in rem jurisdiction is applicable.  For purposes of exercising in rem jurisdiction, the situs of tangible property is straightforward—the property is located where it is physically present, a finding made by the appropriate court.

The Supreme Court in Pennoyer v. Neff, 95 U.S. 714 (1878) disclosed the basics of in rem jurisdiction. It found that courts were empowered to reach both persons and property located within the state's sovereign borders.  In short, the states possess exclusive jurisdiction and sovereignty over persons and property within its territory.

The Pennoyer court also disclosed the due process requirement of notice in these exercises of jurisdiction. More specifically, for in personam jurisdiction over a defendant, the defendant needed to be personally served with process while they were within the territorial borders of the state.  For in rem jurisdiction over property, the court found that  "[s]ubstituted service by publication, or in any other authorized form, may be sufficient to inform parties of the object of proceedings taken where property is once brought under the control of the court by seizure or some equivalent act…" (Pennoyer v. Neff, 95 U.S. 714 , 727 (1878)).

In addition to in rem jurisdiction, two types of quasi in rem jurisdiction may be useful for resolving AI related legal difficulties by providing some compensation for damages.  More specifically, in rem jurisdiction determines the ownership of property in regard to any possible claimant while quasi in rem plaintiffs do not seek to determine the ownership rights in the rem (property at issue).

The first type of quasi in rem jurisdiction implicates a plaintiff asserting their interest in property and seeking to have that interest established against the claim of another designated person.  The second type of the plaintiff is actually not asserting an ownership interest as is the case for in rem and quasi in rem of the first type, but rather is seeking to compel satisfaction of another claim using the property's value.

This second quasi in rem type may be most relevant to AI actions, because the claim against the person that gives rise to the action is not related to the res that provides jurisdiction.  That is, the plaintiff does not contest the property rights of the owner of the res but seeks to obtain the res in satisfaction of some separate claim.

The average cost to develop AI software can range from $10,000 for a simple solution or feature to $200,000 and more for the complex AI.  Securing such an asset may ameliorate or satisfy the damage caused by AI.

Alternatively, a quasi in rem action may motivate an AI owner (located outside the jurisdiction of the court) to voluntarily join the litigation so as to mitigate the damages or be positioned to settle the case.