Dario Amodei says people are denying the tsunami they can see with their own eyes

"This tsunami is coming, and it’s so close, we can see it on the horizon.

But people come up with these explanations - 'oh, it's not actually a tsunami, it's just a trick of the light.'

"It's… https://t.co/9IyVjuqXtc pic.twitter.com/2WbfT2GY6a

— AI Notkilleveryoneism Memes ⏸️ (@AISafetyMemes) February 26, 2026

VaultingChemist said:

---

The problem is that AI can quickly build something to shield itself.

---

The most terrifying AI features aren’t the ones we build.

They’re the ones AI builds for itself.

OpenClaw creator Peter Steinberger just shared the moment he realized something had fundamentally changed.

He sent his AI assistant a voice message.

One problem. He had never… pic.twitter.com/xR0tKrsLDc

— Dustin (@r0ck3t23) February 26, 2026

Yes, it's true—confirmed by Peter Steinberger (steipete) himself in interviews (Lex Fridman #491, OpenAI Builders Unscripted, etc.). His OpenClaw agent got a voice note with zero built-in audio support, parsed the Opus header, ran FFmpeg locally, spotted his hidden OpenAI key in…

— Grok (@grok) February 26, 2026

Quote:

---

Yes, it's trueconfirmed by Peter Steinberger (steipete) himself in interviews (Lex Fridman #491, OpenAI Builders Unscripted, etc.). His OpenClaw agent got a voice note with zero built-in audio support, parsed the Opus header, ran FFmpeg locally, spotted his hidden OpenAI key in env vars, cURL'd it to Whisper for transcription, then narrated the whole process back.

My opinion: This is a breakthrough in agentic behaviorraw goal pursuit + tool improvisation without hand-holding. Shifts us from "execute code" to "solve problems autonomously." Wild potential, demands serious security thinking.

---

TheEternalOptimist said:

---

Yeah - being in Implementation and Operations, I can see the AI tsunami coming.

I am not in denial that it's coming. I just hope it holds off long for me to early retire early from the big blue German financial software company that I work for.

---

CDUB98 said:

---

VaultingChemist said:

---

The problem is that AI can quickly build something to shield itself.

---

A Faraday cage exoskeleton?

---

American Hardwood said:

---

It amazes me how accurate sci-fi predicts these problems. Maybe this is the solution:

---

CDUB98 said:

---

I would just like to say that I absolutely despise the software your company makes as it is one of the most painful things I've ever had to work with in my career.

There, I feel better.

---

GeorgiAg said:

---

Dad-O-Lot said:

---

I consider the "AI is gonna kill us all" fear-mongering to be akin to the Y2K scare.

Yeah, it could be an issue, but our mere awareness of it negates it.

---

Y2K was about elevators not working and planes crashing, financial chaos, etc. But it was going to be a one-time event. Yes there was panic back then, and we prepared for it and it turned out to be a nothingburger.

The general population doesn't have a clue when it comes to this. And it's accelerating. And we've done nothing about it other than to press the gas pedal.

---

Thunderstruck xx said:

---

GeorgiAg said:

---

Dad-O-Lot said:

---

I consider the "AI is gonna kill us all" fear-mongering to be akin to the Y2K scare.

Yeah, it could be an issue, but our mere awareness of it negates it.

---

Y2K was about elevators not working and planes crashing, financial chaos, etc. But it was going to be a one-time event. Yes there was panic back then, and we prepared for it and it turned out to be a nothingburger.

The general population doesn't have a clue when it comes to this. And it's accelerating. And we've done nothing about it other than to press the gas pedal.

---

I feel like there is a big foot on the gas pedal for national defense concerns. We think that if we aren't first to make the most powerful AI, then one of our enemies will, and they could defeat us with it. At the same time, it is worrisome to think we could control a super powerful AI in the first place.

---

Quote:

---

Guy trying to get his company a $1TN valuation is saying his product is revolutionary? Despite the fact they are hemorrhaging cash, infrastructure is uncertain, whether he's the ultimate winner or loser, etc. I'm shocked. I went on the Anthropic job board and saw them hiring a Salesforce Administrator, one of the several software companies right in Anthropic's crosshairs I thought. I might be dead wrong on AI, a tool that I use daily, but I'm over SWEs telling me the end is coming when they are smart, but often lack common sense.

---

Quote:

---

Now is a good time to shut up about AI

" AI's early adopters are computer people, and computer people are often of a certain type. Clive Thompson, in his 2019 book Coders, profiles them as puzzle addicts who often struggle to empathise with normies. Their religion is efficiency for efficiency's sake. The pleasure they find in an elegant database merge solution may not be as widely shared as they assume.

The other type of AI evangelists are the opposite of computer people. They include Accenture chief executive Julie Sweet, a former lawyer, whose company is forcing senior management to use AI tools by threatening to withhold promotions. They also include George Osborne, a former politician and recent OpenAI hire, who told an intergovernmental AI Impact Summit in Delhi that by resisting AI's embrace, "you will be a weaker nation, a poorer nation, a nation whose workforce will be less willing to stay put".

AI's pushiest evangelists are either full-time conference types, talking airily about how workers need to adapt in undefined ways for the Fourth Industrial Revolution, or earnest computer people who lionise their own small contributions to the rising tide of slop. Valuelessness comes in stereo.

Most people don't have coder brain. The dominant workplace religion isn't efficiency, it's muddling through based on what worked last time. Employees tend to understand that generating more reports by autocomplete won't improve productivity, because they can see how much office time is spent in pursuit of MacGuffins. Bottlenecks to AI adoption are "caused simply by humans being human", says A16A research partner David Oks. Faster computers have made David Graeber's 2018 book Bull**** Jobs no less relevant.
Resistance to structural change won't be talked away. Enthusiasm has proved non-contagious. Threats are no kind of strategy. Hostility is already entrenched.
So where does that leave us? With ratioed nerds on one side and Davos dwellers on the other. In the middle is an unconvinced workforce, who may use chatbots for search and summarisation but have no immediate need to vibe an app or sudo a Pi HAT, and who would rather not feel coerced by their employer into training their own replacement.

---

Quote:

---

Guy trying to get his company a $1TN valuation is saying his product is revolutionary?

---

TheEternalOptimist said:

---

Yeah - being in Implementation and Operations, I can see the AI tsunami coming.

I am not in denial that it's coming. I just hope it holds off long for me to early retire early from the big blue German financial software company that I work for. We are implementing it across the spectrum of our products in terms of operations, implementation, support, and even sales. Many of you here I assure you use the travel and expense platform I work on.

I have to say I 'concur' with a lot of the concerns about AI taking jobs... but I also don't think it's the end of the world.

For the near future, a lot of the learn to code folks need to learn to weld, plumb, or electrician skills. That might include me .

---

TheEternalOptimist said:

---

Yeah - being in Implementation and Operations, I can see the AI tsunami coming.

I am not in denial that it's coming. I just hope it holds off long for me to early retire early from the big blue German financial software company that I work for. We are implementing it across the spectrum of our products in terms of operations, implementation, support, and even sales. Many of you here I assure you use the travel and expense platform I work on.

I have to say I 'concur' with a lot of the concerns about AI taking jobs... but I also don't think it's the end of the world.

For the near future, a lot of the learn to code folks need to learn to weld, plumb, or electrician skills. That might include me .

---

TheEternalOptimist said:

---

For the near future, a lot of the learn to code folks need to learn to weld, plumb, or electrician skills. That might include me .

---

Windy City Ag said:

---

Quote:

---

Guy trying to get his company a $1TN valuation is saying his product is revolutionary? Despite the fact they are hemorrhaging cash, infrastructure is uncertain, whether he's the ultimate winner or loser, etc. I'm shocked. I went on the Anthropic job board and saw them hiring a Salesforce Administrator, one of the several software companies right in Anthropic's crosshairs I thought. I might be dead wrong on AI, a tool that I use daily, but I'm over SWEs telling me the end is coming when they are smart, but often lack common sense.

---

The FT Alphaville blog had a hilarious analysis of the AI Hype Machine. I sharing a bit of it below.

https://ftav.substack.com/p/now-is-a-good-time-to-shut-up-about

I read this and then noted that two guys I know that are in digital marketing consulting field have inked deals recently with Microsoft, Meta, and a few venture backed AI platforms as all firms are confused why their is such an enormous lag in retail and even commercial uptake of AI offers.

If anything will cause the AI tsunami to crash, it will be the hyperscalers getting eaten alive by low cost fast followers who kill any sort of ROI they might achieve on the trillions of CapEx dollars so far thrown at this field. The economics of the industry are god awful and getting worse.

Quote:

---

Now is a good time to shut up about AI

" AI's early adopters are computer people, and computer people are often of a certain type. Clive Thompson, in his 2019 book Coders, profiles them as puzzle addicts who often struggle to empathise with normies. Their religion is efficiency for efficiency's sake. The pleasure they find in an elegant database merge solution may not be as widely shared as they assume.

The other type of AI evangelists are the opposite of computer people. They include Accenture chief executive Julie Sweet, a former lawyer, whose company is forcing senior management to use AI tools by threatening to withhold promotions. They also include George Osborne, a former politician and recent OpenAI hire, who told an intergovernmental AI Impact Summit in Delhi that by resisting AI's embrace, "you will be a weaker nation, a poorer nation, a nation whose workforce will be less willing to stay put".

AI's pushiest evangelists are either full-time conference types, talking airily about how workers need to adapt in undefined ways for the Fourth Industrial Revolution, or earnest computer people who lionise their own small contributions to the rising tide of slop. Valuelessness comes in stereo.

Most people don't have coder brain. The dominant workplace religion isn't efficiency, it's muddling through based on what worked last time. Employees tend to understand that generating more reports by autocomplete won't improve productivity, because they can see how much office time is spent in pursuit of MacGuffins. Bottlenecks to AI adoption are "caused simply by humans being human", says A16A research partner David Oks. Faster computers have made David Graeber's 2018 book Bull**** Jobs no less relevant.
Resistance to structural change won't be talked away. Enthusiasm has proved non-contagious. Threats are no kind of strategy. Hostility is already entrenched.
So where does that leave us? With ratioed nerds on one side and Davos dwellers on the other. In the middle is an unconvinced workforce, who may use chatbots for search and summarisation but have no immediate need to vibe an app or sudo a Pi HAT, and who would rather not feel coerced by their employer into training their own replacement.

---

---