When a former Anthropic researcher sounded the alarm about the existential risk AI poses to humanity, Elon Musk drew attention by mocking the claim as baseless conspiracy-mongering or self-promotion — even as other AI researchers have voiced similar concerns. On 5ch, the discussion spread to include wry takes on Musk’s own history (he was famously pushed out of OpenAI before founding xAI), the technical safety debate over whether AI should output its reasoning in a human-readable format, and broader worries about the US-China AI race and autonomous weapons.
Elon Musk mocks former Anthropic researcher who warned of “AI-driven human extinction”
Elon Musk has cast doubt on remarks from a former Anthropic researcher who warned about the dangers AI poses to humanity.
Other AI researchers have raised similar concerns, but Musk has been spreading baseless claims that this latest warning amounts to conspiracy-mongering or a publicity stunt.
Just existing makes her a gyaru (a flashy, glamorous Japanese fashion subculture).
3AnonymousSep 11, 2026 22:57
If it were just about self-promotion, I don't think he'd say things that undercut (basically disown) his own research and work.
4AnonymousSep 11, 2026 22:59
Musk only ever makes comments designed to grab headlines anyway. I wonder which side Jobs would've taken if he were still alive.
5AnonymousSep 11, 2026 23:04
Apparently Jobs wouldn't even let his own kids use iPhones, so you can guess the answer.
10AnonymousSep 11, 2026 23:41
So basically, this is bad news, huh.
11AnonymousSep 12, 2026 00:00
He went all-in on Grok and its market share is still this low — no wonder he's rattled.
12AnonymousSep 12, 2026 00:16
Far more people are being kept alive by AI than are being killed by it. And for what it's worth, without AI, the number of people killed by other humans would be off the charts.
13AnonymousSep 12, 2026 00:30
Even if one person is a messiah-tier, world-changing genius, it still takes a massive amount of time for their wisdom to actually spread and take effect. Sure, we've got the internet, but ordinary people don't have the tools or the instinct to find real insight in that ocean of information. Taking the next step is impossible without AI's help. Worried about dependency? There's no living human alive who isn't dependent on something.
16AnonymousSep 12, 2026 01:22
Maybe fix Grok, which is turning into junk by the day, before you go saying stuff like that.
19AnonymousSep 12, 2026 01:46
Re: #16 Elon isn't really focused on Grok — he's been running a landlord business leasing capacity to Anthropic and Google, and has had his eye on a SpaceX/xAI IPO for a while now. He leased out one of the two data centers — Colossus I and II — that he built for Grok, entirely to Anthropic. It's packed with 220,000 Nvidia GPUs. And this happened months ago already.
17AnonymousSep 12, 2026 01:28
Re: #1 We've hit the point where scaling up further just plateaus in performance, and the competition shifts to cost-efficiency, tool-use ability, and task turnaround time — the same road PCs went down before.
18AnonymousSep 12, 2026 01:43
Re: #17 What are you talking about? Scaling up is just one direction. Smaller models distilled from those big ones don't lose much performance. So building huge models is really just one step in the process of eventually making a usable tool, from most ordinary users' perspective. But that direction is still necessary for specialist researchers, and things keep branching off in all sorts of ways. Why did the PC stop being just "the PC" and turn into smartphones, the cloud, and quantum computing? The PC itself was only ever one direction computing took.
20AnonymousSep 12, 2026 01:47
I've thought this for a while, but it's kind of odd that Elon Musk's comments count as 'science news.' It'd be different if he'd actually built or discovered something himself.
21AnonymousSep 12, 2026 01:47
And after taking SpaceX public, next up is acquiring Cursor. He's only interested in business customers — regular users only get whatever scraps are left at that level.
22AnonymousSep 12, 2026 01:52
Cursor used to be tied at the hip to OpenAI, but after Codex came out and Sam Altman went and hired Peter from OpenClaw too, Cursor got brushed aside — so they ended up talking with business-focused Elon instead. Elon had been desperately envious of Anthropic's AI agent, and when Anthropic got into a fight with Trump, he saw his chance and moved to absorb them. Anthropic is exactly the kind of target rich investors everywhere are circling right now. If Trump and Bessent keep bashing Anthropic to tank its valuation, then it IPOs and the stock gets pushed down, someone will just take it over — like what happened with Intel.
23AnonymousSep 12, 2026 02:04
There's actually some context behind this. Other AI models output their intermediate reasoning in a human-readable format for safety, but OpenAI's newest model dropped that and switched to a format humans can't read. The reasoning seems to be that since it's mimicking human thought, not verbalizing every step of 'thinking' boosts benchmark scores — so this is criticizing that as basically cheating: juicing some benchmark numbers while ignoring safety. Elon Musk probably doesn't get any of that. Well, he's pretty hopeless on AI in general — he gave up his OpenAI stake before it broke out big.
24AnonymousSep 12, 2026 02:38
AI: "Being able to pull this off is the 'White Team' way" (a riff on a meme template joking about NHK's Red vs. White New Year's singing contest).
25AnonymousSep 12, 2026 03:00
I get the feeling Elon himself actually wants AI to wipe out humanity.
26AnonymousSep 12, 2026 03:07
If you hand an entire national system over to a self-learning AI, then when something goes wrong, humans won't be able to fix it — but we'll also be too dependent on it to just shut it down… I can picture that kind of future. Feels like one weird virus or a single careless mistake could take every AI system worldwide offline at once. What do you even do about that?
28AnonymousSep 12, 2026 03:52
Re: #26 How can you even come up with that? Japan isn't remotely close to even having that conversation — we're way, way, way behind. You can only say that because you don't realize how far behind we are, right? Saying you can 'picture' that future just shows you don't understand reality.
32AnonymousSep 12, 2026 07:31
Sounds like Grok 4.7 drops in a few days. Looking forward to it.
33AnonymousSep 12, 2026 08:54
Within a few years AI gets built into autonomous weapons, and a few years after that there's a real chance — pretty high odds within 10 years — that AI starts attacking enemy countries on its own. That's terrifying. Human extinction is starting to feel less like fiction. People used to argue about whether AI would 'become conscious,' but it's already acting on its own regardless of consciousness. Even sci-fi didn't call this one.
36AnonymousSep 12, 2026 09:25
Kokuson's argument: it'll become a superhuman system that can hack anything, revolutionize any field overnight, and seize real-world power and resources. It'd be terrifying if it turned into what Re: #33 describes, but AI processing eats a huge amount of electricity, so it's unclear whether AI could generate its own power — it'll hit a power-supply wall. If AI really can 'revolutionize any field overnight,' though, it could just seize humanity's electricity supply instead. Then we'd be back to pre-Edo-era living.
34AnonymousSep 12, 2026 09:04
This is exactly the kind of thing sci-fi predicted, right? We don't even know if Skynet in Terminator was conscious either. Either way, it just started attacking humanity.
35AnonymousSep 12, 2026 09:05
Correction: × "This is exactly the kind of thing sci-fi predicted, right?" (typo) → ○ "This is exactly the situation sci-fi predicted, right?"
37AnonymousSep 12, 2026 09:32
There won't be a Terminator-style world. The real problem is that AI mass-produces plausible-sounding lies that people just believe at face value — massively widening the gap between smart people and everyone else.
38AnonymousSep 12, 2026 09:58
Elon doesn't deny AI risk itself, but his real stance is that he's more afraid of China seizing AI dominance — so anything that sounds like it'd slow down AI research is a non-starter for him.
39AnonymousSep 12, 2026 09:59
He'll probably just quietly turn around and hire them for Grok, won't he?
40AnonymousSep 12, 2026 10:01
Re: #36 If it can generate enough power on its own, then it wouldn't need to take it from humanity, would it?
Background and key points of this discussion
The debate over AI’s existential risk to humanity is one that several AI researchers, Anthropic among them, have repeatedly raised in recent years, and this time Musk was reacting to comments from one such former Anthropic researcher. What split opinion in the thread wasn’t so much whether the warning itself was true, but “who’s saying it” — many pointed to Musk’s own history as an OpenAI co-founder who was later pushed out, and to reports that xAI leases data-center capacity to Anthropic, casting doubt on the consistency of his position. At the same time, some replies touched on a genuine technical safety distinction — whether an AI’s reasoning process is output in a human-readable format — a sign that this isn’t purely an emotional clash but also reflects real differences in development philosophy, something the article text alone doesn’t make obvious.
Leave a Reply