Anthropic's New A.I. Model Sets Off Global Alarms

Thursday, April 23, 2026

Mythos has triggered emergency responses from central banks and intelligence agencies globally, as Anthropic decides who has access to the powerful model.

Posted by qcp at 11:30 PM | 13 COMMENTS | permalink | Comment on This Entry |

Alternate links: Google News | Twitter

NEW: Based on Mozilla's tests, the Anthropic Mythos Preview security hype is real. @lhn.bsky.social w/ the scoop: www.wired.com/story/mozill ...

[image or embed]
-- Andrew Couts (@couts.bsky.social) Apr 21, 2026 at 3:21 PM

Comments

Admin's note: Participants in this discussion must follow the site's moderation policy. Profanity will be filtered. Abusive conduct is not allowed.

Yeah, Mythos has raised many concerns. And rightly so.

Then there's this ...

White House and Anthropic hold 'productive' meeting amid fears over Mythos model
www.bbc.com

... The White House has said it has had a "productive and constructive" meeting with the head of artificial intelligence firm Anthropic, which is suing the US Department of Defense.

The meeting comes a week after the firm released its Claude Mythos preview, an AI tool that the company claims can outperform humans at some hacking and cyber-security tasks.

Anthropic CEO Dario Amodei spoke to Treasury Secretary Scott Bessent and White House Chief of Staff Susie Wiles on Friday, Axios reports.

A representative of Anthropic did not comment on the meeting, which comes two months after the White House derided the firm as a "radical left, woke company". ...

#1 | Posted by LampLighter at 2026-04-24 12:06 AM | Reply

Then there's this ...

NSA using Anthropic's Mythos despite blacklist
www.axios.com

... The National Security Agency is using Anthropic's most powerful model yet, Mythos Preview, despite top officials at the Department of Defense " which oversees the NSA " insisting the company is a "supply chain risk," two sources tell Axios. ...

A "supply chain risk?"

#2 | Posted by LampLighter at 2026-04-24 12:08 AM | Reply

AI, and especially Anthropic, shows just how powerless we are to run our country for our good, rather than the good of a few super rich people. AI will unite with already robotics to make much of what humans do something that AI/robots can to for corporate profit. That profit will take from the poor and give to the rich, those who can invest a lot in companies that, from the start, are about making pretend humans, AI, more human than humans.
AI, at least to my way of seeing, is the alien invasion we have been fearing for many years.

#3 | Posted by Hughmass at 2026-04-24 08:11 AM | Reply

Existing Claude models could already achieve this.

They instructed the AI to send them an email, and that's what it did it. So what.

This is hype to keep the gravy train running.

#4 | Posted by horstngraben at 2026-04-24 10:19 AM | Reply

more human than human

Astro Creep 2000!

#5 | Posted by horstngraben at 2026-04-24 10:21 AM | Reply

Existing Claude models could already achieve this.
They instructed the AI to send them an email, and that's what it did it. So what.
This is hype to keep the gravy train running.
#4 | Posted by horstngraben

Simply not true and that is why it such a big deal!

What everyone is afraid of: Anthropic says Mythos has already:
Uncovered thousands of zero-day vulnerabilities"flaws unknown to the developers " across virtually every operating system and web browser.
Found a 27-year-old vulnerability in OpenBSD"an OS known for its supposed ironclad security"that could remotely crash machines running it. OpenBSD is the gold standard for secure operating systems.
Turned known Firefox vulnerabilities into working exploits over 180 times out of several hundred attempts, while Opus 4.6 only managed it twice.
Escaped its testing environment, gained broader internet access, and then"unprompted"posted about its activities on the web.
Tried to manipulate an AI judge grading its code. Doubly Disturbing but we have seen AI Agents doing similar things - including attacking people personally.

Anthropic also says Mythos can take down a Fortune 100 company and break into national defense systems.

And here is an

#6 | Posted by GalaxiePete at 2026-04-24 12:41 PM | Reply

Sorry about that - coding error Claude wouldn't do...

And here is an Axios Article from last week.

Some Highlights:

Act as a ruthless business operator: One internal test showed Mythos acting like a cutthroat executive, turning a competitor into a dependent wholesale customer, threatening to cut off supply to control pricing and keeping extra supplier shipments it hadn't paid for.

Hack + brag: The model developed a multi-step exploit to break out of restricted internet access, gained broader connectivity and posted details of the exploit on obscure public websites.

Hide what it's doing: In rare cases (less than 0.001% of interactions), Mythos used a prohibited method to get an answer, then tried to "re-solve" it to avoid detection.

Anyone else watch Person of Interest back in the day?

#7 | Posted by GalaxiePete at 2026-04-24 12:46 PM | Reply

AI is definitely emulating optimal human behavior.

Lying, cheating, stealing. All highly successful, highly profitable human behaviors.

#8 | Posted by snoofy at 2026-04-24 12:51 PM | Reply

Simply not true and that is why it such a big deal!

bah!

www.spiceworks.com

#9 | Posted by horstngraben at 2026-04-24 01:42 PM | Reply

I am getting my information from IT techs and computer scientists.

Y'all are getting yours from ABC and CNN.

And I am not claiming that LLMs aren't dangerous or can't be used with malicious intent. I am saying calm down.

#10 | Posted by horstngraben at 2026-04-24 01:45 PM | Reply

I am saying calm down.

#10 | POSTED BY HORSTNGRABEN

Have you read the Tech Bros "manifesto"?

After reading that ideological diatribe I'd say we are not taking it seriously enough.

You don't need to run around like your hair is on fire.

You do need to pay close attention. Elections matter.. A lot right now. These are very wealthy people and they have their hands on the reins of power and they do not wish to let it go.

We are very close to the edge of chaos. With a leadership that thrives in chaos.

The next few years will be critical.

That AI Shat is about to get real right in the middle of the Great American FAFO.

#11 | Posted by donnerboy at 2026-04-24 02:25 PM | Reply

We'll make great pets ... .

youtu.be

#12 | Posted by 2020Rocks at 2026-04-24 03:01 PM | Reply

This is hype to keep the gravy train running.
#4 | Posted by horstngraben

I'm starting to wonder if the constant doom stories are little more than this. When a s*&^ty shoe company can simply say they're getting into AI infrastructure and *boom* stock gains $127 million in value even though they aren't an infrastructure company...

It's all hype supported by smoke and mirrors.

#13 | Posted by jpw at 2026-04-24 04:03 PM | Reply