AI Take The Wheel: Driver Uses Claude To Hack 1994 Jeep Into Passing Smog Test
A Jeep enthusiast has successfully leveraged artificial intelligence to force a stubborn 1994 Jeep Cherokee to pass California's notoriously strict smog check after failing due to high nitrogen oxide emissions at 25 mph.
When Muhammad Zulali found his classic 1994 Jeep Cherokee failing state emissions tests due to the high NOx levels, he didn't head to a mechanic or pony up for expensive dealer diagnostics. Instead, armed with a cheap Arduino Nano plus Claude and Codex, he has become an overnight viral sensation and masterclass in reverse engineering.
Zulali's Cherokee is from an age before the OBD II interface, which renders the car's vintage ECU completely alien to modern diagnostic tools, let alone capable of accepting firmware updates or wireless patches like modern cars can. Nevertheless, with about $15 in parts, Zulali connected an Arduino Nano directly into the Jeep's Chrysler SCI diagnostic port. From there, he let Claude take the wheel on the software side. Claude quickly figured out the esoteric baud rates and inverted electrical signaling of the car's communication protocol, allowing the system to read live engine sensors and log data during test drives.
After multiple trial-and-error test runs, where tweaks initially made emissions numbers worse before dialing them in, the 32-year old Cherokee finally cleaned up its act enough to pass the smog test. Zulali has since open-sourced the utility code on GitHub under the moniker xj-scanner, letting other gearheads peer deep into legacy automotive brains.
On a broader scale, Zulali's breakthrough may further spark DIYers to experiment using AI with OBD II scanners. Should car manufacturers and state smog regulators be concerned? More importantly, where was AI when my beloved D2 A8 kept failing California smog tests due to a thrown code no one could figure out?
