Famed iPhone, Sony Hacker Says AI Coding Brokers Are a Catastrophe Ready to Occur – Decrypt

George Hotz—the hacker who first cracked the iPhone at age 17 and reverse-engineered the PlayStation 3 earlier than Sony sued him for it—printed a weblog submit Sunday arguing that mass adoption of AI coding brokers will finish in catastrophe, or at the least near it.

“I’m calling it now, the adoption of AI brokers into software program improvement might be one of the crucial pricey errors within the area’s historical past,” Hotz wrote. “Brokers can not program, and it’s taking longer and longer to appreciate that they’ll’t.”

“The output is damaged, however in a method that’s getting tougher and tougher to detect. Which is strictly what you’d count on from an more and more correct statistical mannequin.”

The submit, titled “The Everlasting Sloptember,” arrives 5 days after Andrej Karpathy, considered one of AI’s most distinguished researchers, joined Anthropic’s pre-training staff with the express view that AI brokers have already reworked software program improvement. The 2 males now signify reverse poles of a debate the business hasn’t settled—and each have precise credibility to stake a place.

Hotz did not attain his conclusion from the sidelines. He spent six months utilizing brokers on actual tasks: elements of Tinygrad, his open-source deep studying framework, and an entire firmware reverse-engineering of a USB-PCIe chip. “The agent frontloads all of the progress,” he writes, then arms you what he describes as a slot machine lever—you pull it and hope the ending work will get completed.

It by no means fairly does.

Not about ego

Hotz anticipates the apparent pushback: a programmer who defines a part of his id by means of his craft would naturally resist instruments that threaten to exchange him. He takes the objection critically and dismisses it on the deserves.

“​​I assumed extra in regards to the self price preservation factor. Google’s AFL discovered extra bugs than LLMs and no one felt that method about it. Chess and Go are extra common than ever,” Hotz wrote. And he’s proper within the sense that Chess AI has dominated people for many years and the sport solely grew extra common.

So, his concern is not about being changed. It is about what occurs to code high quality when everyone seems to be utilizing these instruments directly, particularly when Large Tech and Wall Road are consistently pushing for the mass use of those instruments.

“I virtually assume that is some type of psyop to promote brokers,” Hotz argues. “Worry of loss is likely one of the solely methods to make massive firms transfer. Although I feel in that worry they’re making an enormous mistake.”

His central argument is organizational. Excessive performers have tight sufficient suggestions loops to catch agent-generated issues earlier than they ship. They learn the code, spot the errors, and calibrate when to belief the device. “The underside performers will not have that self verify,” he writes—they usually’re those utilizing brokers to provide 10 instances their earlier output. At a big firm, that math produces one thing particular: sooner degradation of common code high quality, masked by sheer quantity.

The end result, he argues, might be “a golden period for buckets and buckets of slop, and a darkish age for gems of high quality.” As a concrete instance, he factors to stories that Apple is pushing AI coding instruments throughout its whole engineering group, then asks merely: “Do you assume macOS will get higher or worse within the subsequent 2 years?”

The place the camps stand

Hotz now locations himself in what he calls the “LeCun/Marcus camp”—referring to Yann LeCun, Meta’s chief AI scientist, and Gary Marcus, a longtime LLM skeptic. Each have argued that language fashions are basically subtle pattern-matchers: They will imitate the distribution of present code, however cannot purpose by means of genuinely new issues from first rules.

Vibe coding—describing what you need in plain language and letting AI generate the implementation—has exploded over the previous yr, and the key labs have positioned agent-based coding as a flagship product. Microsoft reworked GitHub Copilot right into a full agentic system in 2025, with CEO Satya Nadella describing it as a platform-level shift similar to the transfer to cloud.

The pushback to Hotz’s place is not summary. Karpathy, who had been skeptical of brokers earlier in 2025, reversed his place after new mannequin releases and joined Anthropic’s pre-training staff on Could 19—5 days earlier than Hotz printed. He described the following few years on the frontier as “particularly formative.”

Anthropic CEO Dario Amodei stated in Davos that some Anthropic engineers have already stopped writing code themselves, letting fashions deal with it whereas they overview the output. Hotz, for his half, says he tried to do the identical factor and located himself reaching for the guide repair each time.