The WASP AI gun is a functional but extremely crude single-shot .22-caliber pistol created through a conversation with GatGPT, an artificial-intelligence model developed by Cody Wilson’s Defense Distributed. It reportedly fired two of three cartridges during a witnessed demonstration in Austin, Texas.
That makes the WASP interesting. It does not make it a robot-designed Glock killer, an autonomous weapons system or the magical “push button, receive gun” machine suggested by breathless headlines.
The important development is not the awkward plastic pistol. It is an AI system that can help produce firearm designs through ordinary written prompts.
What Is the WASP AI Gun?
The WASP is a mostly 3D-printed, single-shot .22-caliber handgun based partly on an existing design called the Ergo Harlot, which was itself derived from the popular Harlot printable pistol.
According to The Verge’s firsthand report, the prototype combines a pistol barrel with a spring-loaded plastic action. Its appearance includes a bulbous grip, vestigial sights and an incomplete trigger guard.
Those strange features are not secret engineering breakthroughs. They look more like the mechanical equivalent of the extra fingers that plagued early AI-generated pictures.
During the reported demonstration:
- The first cartridge failed to fire after the striker hit its primer.
- The second cartridge fired from a secured test fixture and struck a cardboard target.
- Cody Wilson then fired the WASP by hand.
- That third cartridge fired, although the shot missed high.
Calling the pistol “nonfunctional” would therefore be inaccurate. It discharged live .22-caliber ammunition twice. Calling it reliable, proven or mechanically refined would be equally dishonest.
The test involved three trigger pulls, not a serious endurance evaluation.
Did Artificial Intelligence Really Design the WASP?
Yes, with a substantial qualification.
An AI system contributed to the design, but it did not independently decide to manufacture a firearm. A Defense Distributed employee specifically instructed GatGPT to produce a new, functional design derived from the Ergo Harlot.
Human beings supplied:
- The training material
- The original firearm designs
- The assignment
- The prompts
- The manufacturing equipment
- The commercial components
- The ammunition
- The physical assembly
- The testing and troubleshooting
GatGPT reportedly proposed the WASP name, created design elements and responded to feedback through natural-language conversation. Early changes included addressing a grip configuration that could have endangered the shooter’s finger.
That is meaningfully different from a person simply downloading an established STL file and pressing “print.” It is not equivalent to an AI spontaneously inventing the firearm from first principles.
The most accurate description is an AI-assisted firearm design produced under human direction.
Wilson offered a more ambitious interpretation: Defense Distributed had not merely created an AI-generated gun, but “an AI that can generate guns.”
That claim is considerably more consequential than the WASP itself.
What Is GatGPT?
GatGPT is Defense Distributed’s specialized firearm AI. Reporting describes it as having been trained on material from DEFCAD, Wilson’s extensive online repository of gun and gun-component files.
Instead of functioning as a general chatbot, GatGPT is intended to answer firearm questions, identify components and assist with digital designs.
That specialized knowledge potentially matters more than broad conversational ability. A general AI may know what a firing pin is. A system trained around thousands of firearm files may be better positioned to understand how particular components relate geometrically and mechanically.
The WASP does not prove that GatGPT can produce safe, durable or original firearms on demand. It demonstrates that a person can use conversational AI to participate in the design process and produce something capable of firing live ammunition.
That modest achievement is still important because early technology rarely remains modest.
The first useful consumer 3D printers were slow, finicky machines that produced rough plastic objects. Modern additive manufacturing now produces suppressor components from titanium and Inconel. My US Palm EOS suppressor examination discusses commercially manufactured suppressors using that mature metal-printing technology.
Today’s WASP looks primitive. The design process behind it may not remain primitive.
Is the WASP the First AI-Generated Gun?
Cody Wilson describes it that way, but “first” should be treated as a claim rather than an independently provable historical fact.
Firearm designers already use computer-aided design, simulation, optimization software and automated manufacturing. Hobbyists have also used ordinary AI tools for coding, research and troubleshooting.
What appears novel about the WASP is the deliberate use of a specialized language model to conversationally generate a new printable firearm design, followed by manufacturing and live-fire demonstration.
That is a narrower and more defensible claim than saying artificial intelligence invented its first gun.
The WASP also borrows from an existing family of simple .22-caliber printable pistols. It is not the firearms equivalent of AI independently discovering the locked breech.
Is the WASP Reliable or Safe?
Nothing publicly reported supports calling it either reliable or safe.
A two-for-three firing result is not a reliability test. The witnessed failure involved a struck cartridge that did not discharge. The pistol’s unusual grip, open trigger area and questionable sights also deserve scrutiny.
Jonathan Ferguson, the keeper of firearms and artillery at Britain’s Royal Armouries, told The Verge that the WASP qualified as a viable firearm from a legal perspective because it successfully discharged without a breech explosion. From a technical perspective, however, he characterized it as only marginally functional rather than a sound firearm design.
That distinction is exactly right.
A device can meet the basic definition of a firearm without being something a sensible person should trust near his hands or face. The ability to fire one cartridge establishes lethality, not quality.
No independent testing has established the WASP’s:
- Long-term structural integrity
- Drop safety
- Consistent ignition
- Accuracy
- Heat resistance
- Parts life
- Ammunition compatibility
- Resistance to an out-of-battery discharge
- Performance after repeated firing
A dramatic headline cannot replace those measurements.
Why the WASP AI Gun Matters
The WASP matters because conversational AI could reduce the knowledge barrier between possessing digital firearm files and modifying or generating them.
That does not eliminate the need for judgment. Firearms operate under pressure. A chatbot can confidently produce a bad answer, and plastic shrapnel does not become less painful because the underlying mistake was generated by an impressive model.
AI could nevertheless help experienced designers iterate faster, compare geometries, organize component information and identify possible design conflicts. It could also give inexperienced people unjustified confidence in unsafe designs.
That combination guarantees political attention.
Gun-control organizations will present GatGPT as proof that government must monitor files, printers and private manufacturing. Politicians will propose technological restrictions whose practical effect extends well beyond firearms.
The urge to force every printer to recognize forbidden geometry is particularly dangerous. A system capable of inspecting everything an American prints is also a general-purpose surveillance mechanism sitting inside privately owned equipment.
Computer code and digital design files additionally raise serious First Amendment questions. The government is no longer merely trying to regulate a physical receiver. It increasingly wants authority over information that could describe one.
Private firearm manufacturing has existed in America far longer than desktop printers. New technology changes the tools, not the underlying principle that peaceful citizens should not need government permission to understand or manufacture ordinary arms for lawful personal use.
Federal, state and local laws still matter, and they vary. A person prohibited from possessing firearms does not become eligible merely because a component came from a printer. State restrictions may also regulate privately made firearms, serialization or particular configurations.
The Anti-Gun Internet Has a GatGPT Problem
Major platforms already restrict or suppress lawful firearm material. My examination of the YouTube gun rules for 2026 explains how 3D-printed firearms receive especially restrictive treatment.
GatGPT illustrates why prohibition-by-content-policy cannot eliminate technical knowledge.
Once information can be represented as text, measurements, code or geometry, banning one file does not erase the underlying concepts. An AI model may synthesize information from many sources, revise existing designs or express the same mechanical relationships differently.
That does not mean every generated design deserves praise. Most probably deserve a long inspection from behind a thick barrier.
It means the political fantasy of deleting firearms knowledge from the internet becomes even less realistic.
The Bottom Line
The WASP AI gun is real in the limited sense that the physical prototype successfully fired live .22-caliber cartridges. It is also a crude, derivative and poorly proven design that failed on its first attempted shot during the reported demonstration.
The pistol itself is not revolutionary.
GatGPT may be.
An AI capable of participating in firearm design through ordinary conversation could eventually become useful to competent designers, dangerously persuasive to novices and infuriating to governments that believe controlling information is easier than respecting constitutional rights.
The first AI-assisted printable gun looks ridiculous. So did many first-generation technologies.
Judge the WASP honestly: an awkward proof of concept, not a mature firearm. Then pay attention to what comes after it.
GarandThumb.com was established on May 4, 2012, to document my experiences trying to become a better marksman.