# Good people refuse to do bad things

> Engineer Jacob Coxon resigns from Anthropic after warning AI could kill humans by 2036.
- Title: Good people refuse to do bad things | Shortify
- Summary: Engineer Jacob Coxon resigns from Anthropic after warning AI could kill humans by 2036. Engineer Jacob Coxon resigned from Anthropic to warn AI could kill…
- Keywords: jacob coxon, anthropic, ai safety, self improving superintelligence, wall street journal, frances haugen, us senate, tech workers, smiling exhaustion, ai generated code, productivity, quality
- Source: A journey into a wild pointer — https://carette.xyz/posts/good_people_refuse_to_do_bad_things
- Published: 2026-09-20T13:00:00+00:00
- Read time: 1 min
- Topics: jacob coxon, anthropic, ai safety, self improving superintelligence, wall street journal, frances haugen, us senate, tech workers, smiling exhaustion, ai generated code, productivity, quality
## Jacob Coxon resigns
Jacob Coxon resigned from Anthropic today after three years at OpenAI and Anthropic.

> I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic
## Coxon's warning
Coxon warned AI could kill all humans by 2036, not a marketing stunt.

> The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt
## Coxon's background
Coxon joined Anthropic at 27 for safety reputation, likes solving problems and making the world better.

> Jacob Coxon is not a star. He is not a CEO, not a famous scientist, and does not hold a PhD
## Coxon's decision
Coxon left two months before equity vested to speak about AI safety concerns.

> I no longer have anything to gain by juicing up Anthropic's valuation. I left before any of my equity vested
## Coxon's accusation
Coxon accused Anthropic of cutting corners when racing to self-improving superintelligence.

> if you're under pressure to race, you have to cut corners
## Evan Hubinger confirms
Evan Hubinger, Anthropic's alignment team lead, confirmed AI could kill humans by 2036.

> Jacob is correct here, we really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade
## Similar past case
Frances Haugen resigned from Facebook after finding internal research on Instagram's harm to teens.

> We make body image issues worse for one in three teen girls
## The pattern
Both Coxon and Haugen resigned after companies knew about risks but failed to act.

> The pattern is the point. In both cases the company knew. And, in both cases, someone inside had to decide whether knowing was enough
## The disease
Tech workers face smiling exhaustion as they trade quality for productivity and compensation.

> I can do more, faster, but not better
## Key takeaway

Engineer Jacob Coxon resigned from Anthropic to warn AI could kill humans by 2036.