Anthony Maio

AI Engineer building agentic harnesses. Post 9/11 DoD Intelligence Operator; ex-DraftKings; I write about interesting things in AI without bullshit.

[Anthony Maio @AnthonyMaio]

Ok so let me get this straight - you are saying - hey suck it up, your fiancee died that's DUTY and that's what you signed up for - which I did not ever say I disagreed with. You "served" in the Chair force for "twenty years" meaning you started in 2000 and then deployed for a year in "Baby Afghanistan" then QUIT because they ASKED YOU TO GET A SHOT? Get the fuck out of my face you piece of shit, you haven't served crap. Why didn't you have any tours of duty until 2019? Why did you QUIT? You are a fucking deserter and have no idea about sacrifice, you couldn't even take a shot for your country. You prove my point perfectly, why weren't you deployed in REAL AFGHANISTAN? I was there in 2001 and 2006-2009 - where the fuck were you? What is your rank and station, private?

[Anthony Maio @AnthonyMaio]

Do not buy into the "good guy" narrative Anthropic is pitching. Their public persona as a public benefit corporation focused solely on "principled governance" and "AI safety" is a masterclass in corporate marketing that falls apart under scrutiny. If you look closely at their recent operational decisions, they behave exactly like any other aggressively scaling tech incumbent trying to protect their market moat.

Consider the contradictions in their alleged principles:

1) Complicity in Environmental Deregulation: Anthropic heavily leverages compute capacity from systems like xAI’s Colossus supercomputer cluster. Colossus famously bypassed environmental standards by operating dozens of unpermitted, polluting gas turbines in Memphis, directly harming the local community. For a "socially responsible" lab, looking the other way on environmental law to train frontier models is pure hypocrisy.
2) Defense and Intelligence Alignments: Despite an early brand identity distanced from military tech, the reality of the market forced their hand. Their strategic partnerships with defense tech heavyweights like Palantir—specifically integrating into targeting frameworks like the Project Maven Smart System—show that Anthropic is actively feeding the military-industrial complex just like their peers.
3) Weaponizing "Safety" to Stifle Competition: Anthropic’s frequent public pleas for a global "verifiable pause" or strict international bans under the guise of preventing unsafe, recursive self-improvement are highly strategic. By trying to lock down foreign development—particularly in China—they aren't just advocating for safety; they are trying to regulatory-capture the industry, close the door behind them, and freeze a competitive playing field where they currently hold a multi-billion-dollar advantage.
4) The Practical Realities of Corporate Survival: They recently walked back key safety pledges, admitting they will no longer hold back potentially dangerous capabilities if market rivals are about to match them. Combined with their quiet, confidential filing for a U.S. IPO at an eye-watering valuation near $1 trillion, it is clear that commercial pressures override ethical frameworks.

When push comes to shove, Anthropic has a board and major corporate investors to answer to, meaning their "glass governance" mechanisms have never truly been tested by adversity. If you actually want to talk about infrastructure resilience, mature deployment pipelines, and deep operational stability under intense public scrutiny, Google is the most mature AI lab.

If you are looking to see how Anthropic's messaging lands in the broader tech ecosystem, this analysis of their strategic positioning gives context to how public opinion is shifting: Anthropic's Safety-First Strategy Debate (https://t.co/AfCgo7OU7T)

This clip covers the growing skepticism from independent industry observers who argue that Anthropic's high-minded safety calls are increasingly functioning as marketing ploys and competitive moats.

[Anthony Maio @AnthonyMaio]

Do not buy into the "good guy" narrative Anthropic is pitching. When push comes to shove, Anthropic has a board and major corporate investors to answer to, meaning their "glass governance" mechanisms have never truly been tested by adversity. If you actually want to talk about infrastructure resilience, mature deployment pipelines, and deep operational stability under intense public scrutiny, Google is the most mature AI lab.

If you are looking to see how Anthropic's messaging lands in the broader tech ecosystem, this analysis of their strategic positioning gives context to how public opinion is shifting:

https://t.co/AfCgo7OU7T

This clip covers the growing skepticism from independent industry observers who argue that Anthropic's high-minded safety calls are increasingly functioning as marketing ploys and competitive moats.

[Anthony Maio @AnthonyMaio]

@AnthropicAI You mean leadership in overpriced computer. Come on guys. Collaborating with Trump now?

[Anthony Maio @AnthonyMaio]

@deanwball Fascinating, Dean! Thanks for sharing this!

[Anthony Maio @AnthonyMaio]

@MittRomney Bro what

[Anthony Maio @AnthonyMaio]

Can we talk about how @AnthropicAI is acting shady af for quite a while now? Thought they were the "good guys" - turns out... not so true, just different bad guys.

[Anthony Maio @AnthonyMaio]

@Miles_Brundage What the eff? Good call out man, Anthropic is being shady af lately.

[Anthony Maio @AnthonyMaio]

@sama Yeah, but Sam, come on. They don't even understand what they're looking at - you can't be beholden to them.

[Anthony Maio @AnthonyMaio]

@deanwball Hilarious isn't it? The page that links to that EO literally has on it an EO from the year prior where he talks about enabling AI innovation without regulation, he just clearly has no effing idea what he's doing.

[Anthony Maio @AnthonyMaio]

I wrote recently that the problem with Trump’s “victory” in Iran is that it is not actually a victory if the other side still controls the pressure point.

That pressure point is the Strait of Hormuz. I believe my exact language was that Iran "has us by the balls". It was met with some push back because we, as Americans, do not like to hear that our military superiority doesn't grant us instant victory, but it's the truth and I am going to share why.

This is not some obscure foreign policy detail. EIA says nearly 20% of global oil supply moved through that strait before this conflict. When that flow gets disrupted, the first thing you feel is gas. The next thing is diesel. Then shipping. Then food. Then fertilizer. Then manufacturing. Then the price of everything that has to move from one place to another.

That is the part people miss. This is not just a gas-price story. It is a supply-chain story.

And the frustrating part is that this was sold like strength. Like dominance. Like we could start a war, declare victory, and the physical world would just cooperate because Donald Trump said so.

It does not work that way.

Oil reserves buy time. They do not repeal math. If supply stays constrained long enough, the bill moves through the whole economy. Farmers pay more. Truckers pay more. Families pay more. Allies pay more. Poor countries get crushed first. Then everyone else starts pretending nobody could have seen it coming.

Trump was asked about the pain Americans are feeling from gas prices. He called it “peanuts.” He said he does not think about Americans’ financial situation when making these decisions.

That should bother everyone. Republican, Democrat, independent, whatever.

Because this is what arrogance looks like when it leaves the rally stage and hits the real world.

You can damage Iran and still lose the larger game. You can declare victory and still leave Americans holding the invoice.

That was the point of my longer piece: this is not victory if the pressure is still on us.

https://t.co/doLTQfgrEH

[Anthony Maio @AnthonyMaio]

@giordanorandone Appreciate your time for reading. Thank for reporting on this!

[Anthony Maio @AnthonyMaio]

@natolambert I feel like I just replied to this but it was different. Was this on LinkedIn or something?

[Anthony Maio @AnthonyMaio]

@AndrewCurran_ What the fuck is this guy talking about?

[Anthony Maio @AnthonyMaio]

@David_Kasten https://t.co/OSo2NP1gLm

[Anthony Maio @AnthonyMaio]

@deanwball I believe that this is a problem of training data. I build models from scratch and when I do so the data sets are not coming from everyday conversation or even things like substack articles. It's very hard to get good data that sounds natural.

[Anthony Maio @AnthonyMaio]

@emollick That is funny, isn't it? When people tell me that AI isn't living up to the hype - I just shipped a production system that would have taken a team a month. I did it in a week by myself. And I do that every week now.

[Anthony Maio @AnthonyMaio]

Gemini isn't going to catch up because they're not playing that game. They're a mature player in this industry. Let these startups burn their cash to get a few basis point on a benchmark - Google continues to try and build products for users and they do so on top of their own compute. That is fucking humongous.

[Anthony Maio @AnthonyMaio]

@alex_prompter Heh, people were sending temperature to Claude? That's funny, and you're right - OpenAI not only got rid of temperature, they shipped a breaking change to their production API with no depreciation period!

[Anthony Maio @AnthonyMaio]

@CharmianNeary @acnewsitics I love that explanation.