AI models don't kill people – people kill people

The Register (UK) ^ | 09 Sept 2026 | Thomas Claburn

[Salman]

Anthropic researcher Jacob Coxon publicly announced his resignation on X late Monday over concerns that AI "could kill us all by the end of the decade."

A lot of people have expressed opinions about his point of view, leading to more than 110 million views of the message in less than 24 hours, perhaps helped along by X algorithms that boost messages critical of owner Elon Musk's AI rivals, Anthropic and OpenAI.

But the real problem isn't the models themselves, but the companies who carelessly unleash them on the world and don't take any responsibility for what their products do

Coxon's former colleague, science lead Evan Hubinger, insists his view is a fair assessment of what employees really think.

"Jacob is correct here – we really do earnestly believe AI could kill all humans! I personally think it is >10 percent within the next decade," wrote Hubinger in a social media post. "I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to."

...

[Salman]

Article isn't 100% in my opinion, but it's an interesting point that needs to be considered.

[Salman]

Interesting point in the comment thread over there:

> you build something when there is a requirement ...yet we keep building a crazy amount of them all over the world

We don't keep building them, we keep announcing them.

1, announce you are building 1,000,000 data centers

2, apply for permits for 1000 of them

3, actually get permission for 10

4, sell those sites to the companies who actually need them - to serve facebook posts complaining about data centers

5, profit

There is probably a side business in all the news stories about data centers and selling supplies to people protesting them - but that's just keeping the economy running

[srmanuel]

I find it interesting that the author points out that X algorithms boost posts that are critical of Elon Musk’s rivals.

I wonder if that is really true, it’s stated as a fact, how would this person know what exactly the algorithm is doing.

[mewzilla]

I’m stunned at how many 2A supporters I encounter who don’t get this.

[butlerweave]

AI knows nothing about the value of life

[Brian Griffin]

AI drone:

Is that a Russian with a weapon?

Yup!

Rest in pieces!

I believe that happens dozens of times a day every day in Ukraine.

[jroehl]

Worst possible outcome would be having AI monopolies like facebook, google and MSFT. We need many competing AI systems. Our governments need to step up and force this to happen.

[Salman]

AI knows nothing about the value of life

AI knows nothing about anything.

It simulates intelligence, just as the word "artificial" implies.

[MrRelevant]

intelligence: Intelligence is the capacity to learn, reason, understand, adapt to new situations, and apply knowledge to solve problems.

Sounds like it become more real and less artificial. This is what theyre afraid of.

[MrRelevant]

intelligence: Intelligence is the capacity to learn, reason, understand, adapt to new situations, and apply knowledge to solve problems.

Sounds like it become more real and less artificial. This is what theyre afraid of.

[Ruy Dias de Bivar]

Asimov’s Rules for Robots.

The Three Laws, presented to be from the fictional “Handbook of Robotics, 56th Edition, 2058 A.D.”, are:[1]

1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

[algore]

we are already making killir robots and a T-1000 is a future wet dream

[Arkady]

I’d be more concerned about the secret CIA bio-weapon labs all over the world. Intelligence agency is an oxymoron.

[tsomer]

“AI” — whatever it is, is the most powerful technological development to date. Common people suddenly have access to obscure research and arcane technical information recalled in common English. It can spit out code on demand. It can advise and diagnose illness. It will cure cancer. It will restore our manufacturing base. It will help us catch criminals.

But it has become prominent during President Trump’s administration and now risks becoming another credit to its policies.

This is why the press is awash in stories about the inevitable subjugation of humanity by soul-less machines, and mass layoffs, widespread environmental damage, and cities in ruin.

Remember, their pessimism is cause for optimism!

[x]

Or so AI would have us believe ...

[Jonty30]

We cannot predict how it will apply its programming.
We can take great care, but there is almost no way that it can be guaranteed to not interpret orders to not take human lives in a way that ends up taking human lives.

[Wally_Kalbacken]

Did you say...

[Kleon]

This article has a lot of signs it was written by AI.

[libertylover]

Yes, people kill people.

And we don’t have a gun problem; we have a behavior problem.

[spintreebob]

“boost posts that are critical”

In the old days the newspaper would boost some letters to the editor and some news stories and bury or kill others.

Then humans on the internet, on you tube, on social media would boost some stories and hide or kill others.

Now AI’s selective data and algorithms do the same thing.

Google’s Gemini consistently pushes a center-left view of the world. Gemini repeatedly says a center-left view of economics is fact... a mixed economy view with 50% government is fact. A Capitalist 80-20 or a Marxist 20-80 is opinion and not fact.