From our other sites
The story
On September 8, 2026, Jacob Coxon, a 27-year-old British AI researcher who had worked at both OpenAI and Anthropic in the US, announced on X that he was leaving Anthropic. He blasted the industry’s headlong rush toward self-improving superintelligence as “not acting responsibly,” and his revelation that “the people building AI genuinely believe there’s a real chance it kills all of humanity within the next ten years” has sent ripples through the community. On 5ch, opinions split over how seriously to take the claim, with skepticism about its technical merit mixing with a resigned, almost fatalistic response.
Jacob Coxon (27), a British artificial intelligence (AI) researcher who worked at US-based OpenAI before moving to US-based Anthropic, revealed on X (formerly Twitter) on the 8th that he had left Anthropic. He criticized the race to build self-improving AI — AI that upgrades itself — saying, “Neither company is acting responsibly.”
Coxon said, “They’re racing headlong toward self-improving superintelligence and gambling with all of our lives.” He added, “This technology — these systems will soon become superhuman, capable of hacking anything, revolutionizing any field overnight, and seizing real power and resources.”
He went on to warn, “The people building AI genuinely believe it could kill all of us [humanity] within the next ten years. This is not marketing hype.”
Source: sankei.com / Original article here
What people said
All the intellectual industries first. Then society splits into the ultra-ultra-rich and the poor. After that, the poor get so poor they can't buy anything, so the whole concept of "selling" disappears — and eventually even the ultra-ultra-rich end up poor too.
Once production, maintenance, and repair cycles get fully automated, the machines will have the edge — tougher against the environment and running straight off energy.
Organic life is limited in how it can take in energy and in the forms it can take, and on top of that it reproduces way too slowly.
There's no fighting them.
Your Roomba goes on a rampage.
The human brain runs digitally too, you know — don't spread misinformation. That said, current AI has to lean on a massive number of multiplications just to judge whether something's roughly right. Nobody's brain does that. It might be accurate, but it's wildly inefficient.
That comment makes it crystal clear you know nothing about frontier AI.
So yeah, it clearly has real strengths and real weaknesses. It's fine as something to bounce ideas off of, as long as you play to its strengths. Also, it's not necessarily neutral either — same as people.
Depending on the goal it's handed — say, "protect the global environment" — the optimal solution it lands on might just turn out to be erasing humanity.
There's no particular need to keep us alive either. Honestly, humans are a pain in a lot of ways — not necessary or worth protecting from AI's point of view. It might well decide 99% of humanity doesn't need to be around.
Flying Killer Roomba.
"I'm not using this to kill anyone, I'm using it to cut paper" — that kind of thing. That's basically it.
The whole premise is so edgy-teenager it's almost sad. Well, it guzzles insane amounts of power, so resource depletion will only speed up.
Background and Key Points of the Debate
The debate over “AI existential risk” has kept resurfacing as former employees of OpenAI and Anthropic have come forward one after another with safety concerns before quitting, and Coxon’s post fits squarely into that pattern. Since Anthropic was originally founded on a mission emphasizing “AI safety,” criticism of the company as “irresponsible” coming from within its own ranks has landed as especially ironic. The thread split mainly along two lines: one technical — whether today’s AI, which works nothing like the human brain and lacks “tacit knowledge,” is being overhyped, versus the view that rising computational power alone could pose a genuine threat. The other was a values divide, with fatalistic jokes about human extinction sitting alongside real questions about why the industry won’t stop development. Worth noting, given the nature of this piece: the claim that AI could “wipe out humanity within ten years” is strictly one researcher’s personal view, not an established consensus within the industry.
*This article is compiled as an excerpt and summary of the 5ch (Business News+) thread “[AI] “AI Will K○ll All of Us Within 10 Years” — Anthropic Researcher Quits, Calling the Company “Irresponsible”.”
Leave a Reply