_WHAT’S HAPPENING_

The AI company Anthropic has unveiled Claude Mythos Preview, a powerful new artificial intelligence model capable of identifying software vulnerabilities across major operating systems and web browsers. Citing national security and cybercrime concerns, the company is limiting access to a select group of major technology firms and financial institutions through what it calls Project Glasswing.

According to Anthropic, Mythos has already uncovered previously undetected flaws in widely used systems, including decades-old bugs.

Mythos represents a new step in AI, a model so powerful that it cannot even be released to the public due to concerns that bad actors could use it to easily hack government, banking, and other high-value cybersecurity systems.

_THE FACTS_

  • Anthropic is limiting access to Mythos, providing the model to companies like Apple, Google, Microsoft, Amazon Web Services, CrowdStrike, and Palo Alto Networks.

  • Mythos has reportedly uncovered thousands of vulnerabilities in every major operating system and web browser.

  • The model identified a 27-year-old bug in OpenBSD, a security-focused operating system used in firewalls and routers.

  • It also found a 16-year-old vulnerability in FFmpeg, widely used software that handles audio and video files.

  • Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell met with top bank CEOs in a closed-door meeting to discuss Mythos and emerging cybersecurity risks.

  • Anthropic is in ”ongoing discussions” with US government officials, including the Cybersecurity and Infrastructure Security Agency.

  • The company stated more than 99 percent of vulnerabilities Mythos found remain undisclosed because they have not yet been patched.

  • Anthropic does not plan to make the model generally available but aims to learn how to eventually deploy Mythos-class models at scale.

_OUR TAKE_

An AI model this proficient at identifying vulnerabilities poses a genuine threat if it ends up in the wrong hands, yet offers real promise if deployed responsibly.

The thousands of vulnerabilities already uncovered across major operating systems and web browsers demonstrate capabilities that could shift the entire cybersecurity landscape. Cyberattacks targeting critical infrastructure — hospitals, power grids, government networks — would take on a different character if attackers could automate vulnerability discovery at scale.

But in America’s AI race with China, Anthropic’s latest model marks a significant leap forward and, if properly regulated, could unlock untold capabilities that still feel out of reach today.

See more Upward News in Google: add us as a preferred source

Reply

Avatar

or to participate

Avatar

Don G7 months ago

Anything this powerful will fall into the hands and be used by corrupt politicians and used for illegal purposes. That’s just the way it works. I am terrified at how soon this will happen.

Avatar

Photo Man7 months ago

"Does AI this powerful have more positive or negative potential?" With all the evil intended people anymore in this world, you know someone will do and pay anything to accomplish access to Mythos in order to create pure costly chaos upon society.

Avatar

Karl Hermansen7 months ago

A-I is a bridge too far. It's sort of like the atomic bomb. Once we see the devastation, if we survive, we will swear never to use it again. But it will be there, hanging over our heads, if we still have them. You can never go home again.

Avatar

James G.7 months ago

That's true. However, I think the Navy might have a different view on nuclear power. Also, where did the electricity come from that powers your devices? The next question, that I believe you're proposing is, what's the AI equivalent of Hiroshima? I think you're right! There will be a fallout, then the rest "will be there hanging over our heads." Hmmm, you bring up a good point.

Avatar

Michael H.7 months ago

I agree with your commentary. Tools THIS powerful should not be in the hands of just anyone. I also believe access needs to be limited to government as well. In the hands of ANYONE wishing to use it for nefarious ways, it could be catastrophic. That includes our so called "public servants" in government. This kind of tech needs to be kept on an extremely short leash.

Avatar

Leon Powell7 months ago

I sure wouldn't trust a government with that kind of power. Power corrupts absolute power absolutely corrupts.

Avatar

Todd MacQueen 7 months ago

Like with any tool, how it is used, for evil or for good, comes down to the ones using it. A hammer, gun, car, or pencil can be used for either. What makes the difference is the moral and ethical condition of the one using it.

This level of AI is right on track for Super AI to become a thing in the next year. Maybe two. Whoever gets it ensures their dominance and independence as a nation for the foreseeable future. Those who are behind in the race will do whatever is necessary to not be left behind. First they will try to steal the technology. If that doesn't work they will sabotage the leading AI labs to buy themselves time (while also trying to steal it). If that doesn't work, expect overt action. This is the new Manhattan Project. It will shape the world of the near future.

Avatar

Shaun Hurrie7 months ago

In the hands of mankind, who is basically bad not good, I would say more negative than positive. Thankfully there is a GENUINE Intelligence actually running the show...and He is basically GOOD!😌🙏

Avatar

Guy M.7 months ago(edited)

You took the words out of my mouth, Shaun. We're always being fed the line that humans are basically good. If that were true, God most likely wouldn't have needed to send Jesus to Earth to die a horrible death for our sins. Humans are basically bad, just as you say, and history proves this over and over again. AI will be used for both good and bad, but the bad will far outweigh the good. It's guaranteed. It's our free will nature, and it's getting worse by the day. Only Jesus saves. He's knocking at the door. Let Him in while you still can, and enjoy a life you didn't know was possible.

Avatar

Jorge Rios7 months ago

Like most things that the human race is involved with, the default or the easy road is the one usualy taken and leads to bad or negative things.

Avatar

Leon Powell7 months ago

No one can say. New technology is always a double edged sword, it comes down to who wields it and how.

Avatar

Mike Fankhauser7 months ago

Who do you trust? Powerful AI just becomes more powerful and is used as a destructive tool in the wrong hands. There will become a time when AI will be unstoppable and we will all suffer the consequences.

Avatar

Stuart Denny7 months ago

If the secrets of the Manhattan project could not be kept, how likely is it that a publicly deployed software will remain in the right hands no matter what the controls and intent are?

Avatar

Fred Scott7 months ago

Most definitely!

Avatar

Glenn Meyer7 months ago

It’s not, if it gets in the wrong hands, it’s when it gets in the wrong hands. Why would anyone think it can be protected any more than all the other data that is being hacked everyday.

Avatar

Guy M.7 months ago

The love of money is the root of all evil. It's guaranteed that there are lovers of money employed by Anthropic who will be unable to ignore the vast sums of money that will be offered to them by criminal entities who are salivating like Pavlov's dogs at the very real and present capability of being able to "...hack government, banking, and other high-value cybersecurity systems." AI has evil written all over it. See the plethora of signposts ahead -- we're entering "The Irreversible Zone"!

Avatar

Jim Powers7 months ago

AI is a valuable tool when used as it was designed for. I use it to write code once or twice a week now when I can't figure it out. I have used it to organize thoughts for presentations. The problem is bad actors who only desire to cause harm. If we could keep them out, that would be great.

Avatar

Dave R.7 months ago

negative. Like so many things these days, the importance of releasing it safely and with regulation is a "must." I just dont have faith this will be the case.

Avatar

David Hooten7 months ago

Positive

Avatar

Selene Kitchens7 months ago

Depending on ai for anything will bring USA down from the inside Selene Humans arise and do it yourself or let the kids go down with the ship

Avatar

Kate Devaney7 months ago

It does great harm faster, more efficiently, and cheaper, than ever before.

Avatar

Terry McCall7 months ago

There are lots of weapons out there hanging over our heads that are also used for the good of civilization. We can't go back. If it's the end of the world then, it is but we can't put the genie back in the bottle. We have to figure out how to use the genie for good and stop the bad.

Avatar

Jennifer Vaughn7 months ago

There have been so many "woke" & America-hating individuals embedded in institutions and companies who have been willing to leak information for profit. So it is scary imagining that someone is going to just leak this AI tech to our enemies.

Avatar

James G.7 months ago

The key phrase in the above article is; "if properly regulated." And that's the answer to the question!!!

The AI we know today does what it's told. AI receives it's instructions from humans. Humans have a long history of instability. Some can be trusted, others cannot.

The answer to this question, lives in the people behind it. History is proof that our civilization hasn't answered it yet.

Find the answer to this question, successfully implement it, and you'll never have to worry about money again! Or govt's, police, taxes, crime, guns, cheaters, etc. etc.

Imagine a world, where the entire civilization get's behind and follows the 10 commandments, without question, backlash or worry!

Avatar

Bonnie Hissom7 months ago

AI this powerful can only eventually have a huge negative impact. In the wrong hands, who knows what can happen?

Avatar

Bernie Peer7 months ago

Who knows, but probably not a good idea to rush something so unknown. Salesman always wants to rush

Keep reading

View more
caret-right