The original source for this is an Economist article that is paywalled. Truth or not?

Quote:

---

The Economist titled "Donald Trump's blocking of Anthropic is capricious and chaotic."

The publication reports that on June 11, Senate Intelligence Committee Vice Chair Sen. Mark Warner (D-Va.) stated that Gen. Joshua Rudd (head of the NSA and U.S. Cyber Command) had told him that Anthropic's Mythos AI "broke into almost all of our classified systems, not in weeks, but in hours." This was in the context of authorized testing/red-teaming exercises highlighting the model's offensive cybersecurity capabilities.

---

NSA SHOCKER: MYTHOS BREACHED NEARLY ALL CLASSIFIED SYSTEMS IN HOURS, NOT WEEKS

NSA CONFIRMS MYTHOS PENETRATED MOST CLASSIFIED NETWORKS WITHIN HOURS

CYBER EARTHQUAKE: MYTHOS CRACKED TOP U.S. SECRET SYSTEMS IN HOURS

— First Squawk (@FirstSquawk) June 21, 2026

marcel ledbetter said:

---

Can we sic it on China?

---

My first thought.

Yeah, not getting that at all.

Unforgiven94 said:

---

Who says we haven't already?

---

Honestly I assume we have.

How have we not used it to trash Iran? I hope we have in ways that we just don't know about.

This now widely circulated claim is based on a line I wrote last week (https://t.co/cygMlroojn). I accurately quoted Mark Warner, vice chair of the Senate intelligence committee, saying that the NSA chief had told him that Mythos “broke into almost all of our classified systems,… https://t.co/CXewtuT5TF

— Shashank Joshi (@shashj) June 21, 2026

It's gonna be interesting living in a world without secrets. Only liars and crooks should worry. I'll be in my garden.

CrackerJackAg said:

---

marcel ledbetter said:

---

Can we sic it on China?

---

My first thought.

---

And steal back what they stole from us years ago?

doubledog said:

---

Logos Stick said:

---

The original source for this is an Economist article that is paywalled. Truth or not?

Quote:

---

The Economist titled "Donald Trump's blocking of Anthropic is capricious and chaotic."

The publication reports that on June 11, Senate Intelligence Committee Vice Chair Sen. Mark Warner (D-Va.) stated that Gen. Joshua Rudd (head of the NSA and U.S. Cyber Command) had told him that Anthropic's Mythos AI "broke into almost all of our classified systems, not in weeks, but in hours." This was in the context of authorized testing/red-teaming exercises highlighting the model's offensive cybersecurity capabilities.

---

---

Let us start with the obvious : I would not believe Mark Warner if he told me water was wet.

---

Ok, we have our first "it's BS" post.

Even the author of the original article is saying it is being taken out of context.

He said this:

"But it would be a mistake to read that literally, I think. It surely depends on using Mythos alongside other tools under very particular conditions."


He thinks?!?! It surely depends?!?!

The quote was not taken out of context. The author simply decided to editorialize, based on what is anybody's guess.

Logos Stick said:

---

He said this:

"But it would be a mistake to read that literally, I think. It surely depends on using Mythos alongside other tools under very particular conditions."


He thinks?!?! It surely depends?!?!

The quote was not taken out of context. The author simply decided to editorialize, based on what is anybody's guess.

---

Sounds to me like he realized he caused a panic and it's coming back to him and/or Warner feeling heat and he's getting leaned on to try and caveat to mitigate Warners traitorous talk for making this public.

The author of an article about how ridiculous the ITAR move was says you're reading too much into what one guy said he heard from someone else. I know F16 is predisposed to panic about anything AI, but it's pretty funny how little you guys understand and how much you're willing to jump to conclusions based on a tweet (without even reading the underlying article first).

It was a mistake to pull Fable. Hope they reinstate it soon. Complete overreaction

How little we know about AI? Educate us then, oh great AI guru!

"one guy said he heard from someone else."

This "one guy" is a member of the Senate Intelligence committee. The "someone else" is the head of the NSA!

The author has an opinion about what Gen. Joshua Rudd (head of the NSA and U.S. Cyber Command) said and you, without anything else but the author's opinion, decided that Mythos did not do what he said it did.

He didn't explain what it did, where it started, how it was prompted, how many parallel processes it used, etc. That's exactly how we know you don't understand the statement.

And the author got it from someone who said he heard it from Rudd. Reading comprehension is key. This is exactly how the Russia lie got spread. I thought we had learned our lesson to check sources and make sure it isn't a circular argument.

Pinochet said:

---

He didn't explain what it did, where it started, how it was prompted, how many parallel processes it used, etc. That's exactly how we know you don't understand the statement.

And the author got it from someone who said he heard it from Rudd. Reading comprehension is key. This is exactly how the Russia lie got spread. I thought we had learned our lesson to check sources and make sure it isn't a circular argument.

---

He clearly said what it did!

So it's a nothing burger because the prompts used, the hardware it ran on, etc... was something unique to this trial and any other LLM or even non-LLM techniques could have done the same thing in the same amount of time had they gotten the benefit of the test methods that they used with Mythos?! That's your assertion?! I'm sure OPUS 4.8 - the previous Anthropic model - (and other LLMs) were also tested but did not perform as well because they were handicapped, right?! :-)

LOL. The "prompts used" argument is especially laughable. The fact that you believe the testbed was materially different is exactly how we know you don't understand how these things work.

Logos Stick said:

---

Pinochet said:

---

He didn't explain what it did, where it started, how it was prompted, how many parallel processes it used, etc. That's exactly how we know you don't understand the statement.

And the author got it from someone who said he heard it from Rudd. Reading comprehension is key. This is exactly how the Russia lie got spread. I thought we had learned our lesson to check sources and make sure it isn't a circular argument.

---

He clearly said what it did!

So it's a nothing burger because the prompts used, the hardware it ran on, etc... was something unique to this trial and any other LLM or even non-LLM techniques could have done the same thing in the same amount of time?! That's your assertion! I'm sure OPUS 4.8 - the previous Anthropic model - (and other LLMs) were also tested but did not perform as well because they were handicapped! :-)

LOL. The "prompts used" argument is especially laughable! The fact that you believe the testbed was materially different is exactly how we know you don't understand how these things work.

---

Tell us how they work then big boy. We are waiting.

twelve12twelve said:

---

Logos Stick said:

---

Pinochet said:

---

He didn't explain what it did, where it started, how it was prompted, how many parallel processes it used, etc. That's exactly how we know you don't understand the statement.

And the author got it from someone who said he heard it from Rudd. Reading comprehension is key. This is exactly how the Russia lie got spread. I thought we had learned our lesson to check sources and make sure it isn't a circular argument.

---

He clearly said what it did!

So it's a nothing burger because the prompts used, the hardware it ran on, etc... was something unique to this trial and any other LLM or even non-LLM techniques could have done the same thing in the same amount of time?! That's your assertion! I'm sure OPUS 4.8 - the previous Anthropic model - (and other LLMs) were also tested but did not perform as well because they were handicapped! :-)

LOL. The "prompts used" argument is especially laughable! The fact that you believe the testbed was materially different is exactly how we know you don't understand how these things work.

---

Tell us how they work then big boy. We are waiting.

---

The internet is a series of tubes....

unimboti nkum said:

---

It's wild that we're intentionally summoning a demon

---

Truth. Why we've chosen to go down the path of our own destruction is yet another failing of the human being.

marcel ledbetter said:

---

Can we sic it on China?

---

We probably already did

YouBet said:

---

And also wild that the Senate Intelligence Committee Leader is telling the world our most top flight security group is helpless against AI.

Really f'ing responsible.

---

That was my first thought, too.

Well, my second thought. My first thought was that I'd really love a cheeseburger right now