california introduced AB 2047 this week. the "Firearm Printing Prevention Act." it requires every 3D printer sold in the state to run DOJ-certified software that scans your files before printing and refuses to make anything that looks like a gun part. if you tamper with the software, that's a misdemeanor. if you sell a non-compliant printer, that's $25,000 per unit.
washington has a similar bill. new york has one that covers CNC mills too. three states, a quarter of the US population, moving in lockstep.
the trigger was Luigi Mangione. december 2024, a UnitedHealthcare CEO shot with a partially 3D-printed pistol. ghost gun seizures in new york went from 17 in 2018 to 438 in 2024. the numbers are real. the fear is real. the solution is not.
here's the problem. a 3D printer doesn't know what it's making. it reads coordinates and moves a hot nozzle. asking it to detect a gun is like asking a pen to detect a threat letter. the geometry of a trigger guard looks like a bracket. a barrel looks like a tube. a receiver looks like a box with holes. there are thousands of legitimate objects with those same shapes, and infinite ways to design a gun that doesn't match any known pattern.
currency has a solution for this. the EURion constellation is a pattern of dots printed on banknotes that photocopiers can detect. it works because currency is standardized. there are a handful of bill designs in circulation and they all carry the same marker. guns are the opposite. anyone can design one from scratch. there's no constellation to look for.
the only company selling detection software right now is a spanish outfit called Print&Go. their product is called "3D GUN'T." it compares files against a database of known gun designs and uses a camera to watch the print in progress. the CEO of Glowforge, a laser cutter company that got the same kind of pressure from Manhattan's DA, said it plainly: "software not only doesn't exist, it can't exist because you can't look at physical pieces and determine conclusively whether or not it's going to turn into something dangerous."
bruce schneier compared it to DRM. cory doctorow predicted it in 2011, in a talk called "the coming war on general-purpose computing." the argument is simple: every time you try to stop a general-purpose device from doing a specific thing, you have to add surveillance and control mechanisms that make the device worse at everything else it does. DRM made it illegal to rip CDs you owned. region locks prevented you from playing DVDs you bought. locked bootloaders turned your phone into a kiosk. none of these stopped piracy, because the people motivated to circumvent them always could.
3D printers run open-source firmware. marlin. klipper. the slicer software that converts a 3D model into printer instructions is open-source too. cura. prusaslicer. any blocking at the hardware level can be bypassed by flashing new firmware, which takes about ten minutes. the bill effectively requires manufacturers to lock their devices down, which means killing the open-source ecosystem that made desktop 3D printing possible in the first place.
the real play is market pressure. california, new york, and washington together are big enough to force manufacturers worldwide to either build separate product lines or just put the blocking software in everything. the same dynamic that made california's auto emissions standards national. but cars have closed ECUs. printers have USB ports and UART headers and github repos full of firmware source code. it's a different kind of machine.
what bothers me isn't that legislators want to stop ghost guns. ghost guns are a real problem. what bothers me is the pattern. a bad thing happens. the response is to make a general-purpose tool less general-purpose. the tool gets surveillance bolted onto it. the surveillance doesn't catch the bad actors, because they're motivated enough to work around it. but it does catch the teacher printing parts for a robotics club, and the hobbyist printing replacement hinges for kitchen cabinets, and the small manufacturer prototyping components. the tool gets worse for everyone except the people it was supposed to stop.
this is the template. it was the template for DRM, for age verification that requires face scans, for chat monitoring proposals in the EU, for content filters on social media. regulate the capability, not the intent. surveil the tool, not the act. it never works and it always costs something.
AB 2047 sets up a two-year bureaucracy. the DOJ has to define detection standards by july 2027. certify algorithms by january 2028. publish a roster of compliant printers by september 2028. sales ban by march 2029. three years from now, california will have a list of DOJ-approved 3D printers and a misdemeanor for anyone who makes their own printer do what they want.
making a gun without a serial number is already illegal in california. has been since 2016. manufacturing firearms without a license has been illegal since 2024. distributing CAD files for printing guns has been illegal since 2025. the act is already criminal. AB 2047 doesn't make the act more criminal. it makes the machine a suspect.
somewhere in a garage in sacramento, a kid is printing a gear train for a school project. the printer pauses, scans the file, decides it doesn't look like a gun, and continues. the kid doesn't notice. nothing happened. that's the best case. the worst case is the same kid, same project, and the algorithm flags a housing component because it has the wrong ratio of cylinders to rectangular voids. print refused. explain that to a 14-year-old.
a spring-shaped part has no way of revealing its intended use. neither does a tube, a frame, or a box with carefully placed holes. the geometry is innocent until someone assembles it with intent. you can't regulate intent by interrogating geometry. you can only pretend to.