The AI Journal

New York’s Digital Fair Repair Act was signed into law in December 2022 as the first electronics right-to-repair statute in the United States, and it arrived with an asterisk its drafters had spent years trying to avoid. In the weeks before Governor Kathy Hochul signed it, the bill gained two amendments: manufacturers could satisfy the law by selling pre-assembled bundles of parts rather than individual components “where the risk of improper installation heightens the risk of injury,” and they were not required to supply passwords or tools that would let a device bypass its own security checks. Advocates who had tracked the bill for years called it a win with an asterisk attached. It took the next two years of ordinary screen and battery replacements to show what the asterisk actually did.

What it protected, without naming it, was parts pairing. On a growing share of consumer electronics, an individual component — a screen, a battery, a camera module, a biometric sensor — carries a serial number that the device checks, at the moment of installation, against the manufacturer’s own authorisation database. Install a screen pulled from an otherwise identical device of the same model, and the replacement can be physically and electrically correct while the device still throws a persistent security warning, disables automatic colour calibration, or refuses to report battery health accurately. New York’s statute obliges a manufacturer to sell the part and the manual for fitting it. It says nothing about whether the part, once fitted, is allowed to work as the original did, and the injury-risk exception gives a manufacturer a label under which almost any serialised component can be classified as an assembly rather than a part. A repair shop can now buy a genuine battery from the manufacturer, as the law requires, and still be handed a version that will not clear the same authorisation check as the one it replaced.

Oregon’s right-to-repair law, signed in March 2024 and effective the following January, took a narrower approach and produced a sharper reaction. It did not expand what parts, tools or documentation a manufacturer must provide beyond what New York, Minnesota and California already required. It added one further constraint: a manufacturer may not use software to prevent an owner or an independent repair provider from installing a genuine or compatible replacement part once it has been fitted correctly. Apple, which had endorsed California’s broader access bill the previous year, lobbied specifically against Oregon’s pairing clause. The same company drew opposite conclusions about two statutes from the same legislative family, and the only variable that changed between them was which lever the law pulled. A broad mandate to supply parts was tolerable. A narrow rule about what the installed part is allowed to do was not.

The Federal Trade Commission’s 2021 report to Congress, Nixing the Fix, had already supplied the empirical grounds for treating that reaction as revealing rather than incidental. Surveying manufacturer justifications for repair restrictions across the consumer electronics, agricultural and medical-device sectors, the Commission found scant evidence that restrictions on independent repair protect consumers from unsafe repairs, guard against data breaches, or serve any purpose beyond channelling repair revenue back to the manufacturer or its authorised network. But the report treated repair restriction as one undifferentiated category — parts access, diagnostic access, documentation access and software locks bundled together — and the broad-mandate statutes it helped inspire inherited that lack of differentiation. Colorado’s 2023 agricultural equipment act, aimed squarely at John Deere’s long-standing repair restrictions, requires manufacturers to supply parts, embedded software, firmware, tools and documentation to owners and independent shops. It says nothing, any more than New York’s law does, about whether an installed part is permitted to be recognised once it is in the machine.

Aaron Perzanowski’s account of the underlying property question explains why that gap matters more than it looks. His 2021 study of ownership in the digital economy documents how manufacturers have layered contract terms, software locks and design choices on top of physical goods until ownership of the object no longer guarantees the use of it. Parts pairing is the cleanest instance of that argument available, because it removes even the residual case a manufacturer might once have made about counterfeit or substandard components: the part in a paired-out repair is not counterfeit, not substandard and not incorrectly installed. It is the genuine article, correctly fitted, and the device still treats it as foreign until a remote server says otherwise. A broad access mandate cannot reach that failure, because the failure occurs after access has already been granted.

The European Union’s Right to Repair Directive, adopted in July 2024 with a transposition deadline of July 2026, arrives at a similar diagnosis by a different route. Rather than naming parts pairing, it bars manufacturers from using “contractual, hardware, software or other means” as an obstacle to repair without a “legitimate and objective justification” — a general test built to catch exactly the behaviour Oregon’s statute names directly, without New York’s assembly exception available as an escape hatch.

The strongest objection to generalising Oregon’s rule is that some pairing serves a function the New York carve-out was gesturing at honestly, even if it drew the boundary too wide. Counterfeit lithium-ion battery cells are a documented fire hazard, not a manufactured pretext: work by Marchiori and Conti, presented at the 2023 ACM Conference on Computer and Communications Security, treats battery authentication as a genuine defence against cells prone to thermal runaway, and a manufacturer that disables fast charging on an unauthenticated battery is not obviously behaving anticompetitively. That objection is real, and it does not rescue New York’s version of the exception, which lets a manufacturer invoke injury risk for a camera module or a biometric sensor with no comparable hazard behind it. Oregon’s own statute already draws the narrower distinction, exempting pairing tied to a specific, demonstrable safety function rather than any component a manufacturer chooses to designate. The reform this argument supports is not a categorical ban on pairing everywhere; it is a shifted burden, applied part class by part class, in which a manufacturer must show the hazard a given lock addresses rather than asserting one by label. A battery plausibly clears that bar. A screen does not.

What the New York–Oregon comparison actually shows is not that comprehensive repair legislation has failed; it is that comprehensiveness was never where the leverage sat. A bill that hands over a warehouse of parts and manuals can be neutralised by four words inserted in a late amendment. A bill that does nothing but forbid one specific behaviour, part by part, made the same manufacturer lobby against a single state for the first time in a decade of fighting over the broader mandate it had already learned to live with.