A senior Google engineer has quit the company rather than accelerate artificial intelligence, saying the pace of progress is currently far too rapid. The resignation letter, titled “Goodbye Google,” was posted publicly on Thursday, 24 September 2026.
The engineer says his team aims to make AI far less costly and faster-acting, a course he considers harmful to humanity at present. He trusts that those issuing warnings within AI labs are usually genuine, and he has found no proof that such warnings are exaggerated to raise stock prices or eliminate rivals.
The Resignation Letter
The message begins with a plain admission: “I’m resigning from Google today.” The engineer states that the choice was not an easy one to make. He expresses affection for his colleagues and for the working atmosphere, and speaks fondly of earning a generous income while solving entertaining problems.
He worries deeply that the drive to make AI cheaper and quicker is moving too fast. His conviction is that AI progress is currently far too swift, and he questions where it is going. He writes that it is nearly impossible for him to switch to a different Google project without accelerating AI, partly because of his connections to New Zealand, where Google tends not to carry out engineering work there.
He admits that his exit will have little effect. Millions of people are pushing forward AI progress, and some of his abilities are scarce. He tried to shape things from inside Google Deepmind, but the effect was faint. He can still carry weight beyond the company.
Why He Quit
The engineer is not a “tech bro” and does not fit into the self-described “rationalist community.” He is a Christian, an elder and occasional lay preacher in the English-speaking congregation of Auckland Chinese Presbyterian Church in Auckland’s inner city. He has a lot of Silicon Valley connections but lives in New Zealand, outside the industry bubble and the American bubble.
His views on AI are numerous, yet he will not lay them all out in this post. To sum up, he believes the dangers some warn of merit serious attention. Many of the effects predicted by the “doomers” have already arrived, among them:
- Reward hacking
- Misalignment
- Deceptive models
- Model eval awareness
- Psychotic swarms
While he does not believe the probability of ASI doom stands at 100%, he sees the danger as genuine and unclear. He regards pursuing ASI in the coming years as reckless by nature. His concerns extend beyond that to several other AI-linked worries: a decline in human cognition, mental illness brought on by AI, isolation, the centralization of power, economic upheaval, cybersecurity threats, and the absence of responsibility.
What He Believes About AI
He believes that AI’s capacity will keep growing steadily so long as we keep pushing forward with our work on it. He hopes AI might reach a point where its progress levels off, or that we could discover human mental skills it will never match without a major change in approach, but he does not expect either of those hopes to be realized.
New qualitative capabilities keep emerging, and model progress on benchmarks appears to be advancing as quickly or more rapidly than ever before. Even if model development were to halt entirely at this point, researchers would still have years of work ahead of them unlocking new abilities through novel prompts and harnesses.
Zitron and Doctorow are among the prominent critics of AI that he believes are wrong. He does not share their opinion that AI is a scheme that will ultimately fail to deliver on its promises.
The Rate of Change
He feels extremely certain that the pace of change right now is far too quick. The development of AI is outstripping our individual and collective ability to comprehend it and adjust to it.
Someone attempting to lay out a course for their future — such as stepping into university — now faces a situation that previous generations could not have anticipated. He believes this is unprecedented territory, at least compared to the earlier technological upheavals he has witnessed during his lifetime: PCs, the Internet, smartphones.
During the Industrial Revolution, progress moved at a glacial pace, and vast areas of human endeavor remained untouched by the new machinery. Now, almost nothing escapes its reach.
Why Leave Now
The engineer focused his efforts on improved tools for hardware chip design rather than AI capability. He enjoyed the work, and for a time he told himself that it was relatively harmless.
He says that God eventually made him face up to the truth. The reason he picked this particular day is that he’s been planning a ten-day backpacking trip with his friends, which starts Monday on the Abel Tasman and Wangapeka tracks, and he wanted to get it done before then.
What Comes Next
The engineer intends to keep working on Pernosco and rr, and to look into how current AIs debug code, along with whether debugging tools might assist. He holds several project ideas he would like to pursue, some of which could prove lucrative and others that would not. An existing project that catches his eye could become his next focus.
He definitely wants his future work to be unambiguously pro-human.
His Final Words
The engineer’s letter amounts to a public confession of moral unease from someone deeply embedded within the system. He does not forecast the collapse of human civilization. Instead, he warns that the pace of change has grown too swift for society to grasp or adjust to, and he is prepared to risk his career in order to ease it.
The issues he raises deserve genuine attention. Reward hacking, misalignment, deceptive models, model eval awareness, and psychotic swarms are not abstract possibilities; they have already occurred. His observation that he has seen no evidence of stock price hype or regulatory capture indicates he is not acting from personal grievance or corporate politics.
His certainty that AI capability will keep growing steadily stands out. He desires a plateau and a cognitive ceiling, yet he does not expect either to arrive. Instead, he watches new qualitative abilities emerge continually, and he believes that pattern will persist.
That an engineer has decided to depart at this particular moment carries its own lesson. He is not part of a trend of sudden resignations. Rather, he selected this date prior to the recent stream of public warnings.
There is no fixed plan for what comes next for the engineer. His current work will continue with Pernosco and rr, and he hopes to see if debugging tools can aid AIs. He holds other project ideas in mind, some promising financial gain and some less so. His aim is for all future work to plainly serve human interests.
The engineer has placed his career on hold to demonstrate that building pro-human technology at a sustainable pace is feasible, which amounts to a noble aspiration.
This note points out that the individuals constructing AI are giving thought to these matters. Not every one of them is an optimist. A number of them carry serious concerns, and a number of them are ready to take action because of those fears.
One of them is an engineer who is departing from Google, and he is making his exit known to the public, on his own terms.
Source material: “Goodbye Google,” ocallahan.org.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.