'Terrifying warning sign': Anthropic delays AI model over security concerns

Anthropic has decided not to release its latest AI 

model uh called clawed mythos to the full public. Now this version is apparently so advanced it 

can find vulnerabilities in a huge variety of software applications which is both a great 

great advancement potentially for improving cyber security and also a huge danger if used by 

criminals or others to hack into systems. For now, Anthropic is only making this system available 

to some of the biggest tech companies so they can test it and improve their own systems. The 

fear of course if it fall into the falls into the wrong hand, cyber criminals, spies, the fallout 

could be catastrophic. My next guest, Tom Freeman, lays out those concerns in his latest New York 

Times op-ed with the headline, Anthropics: Restraint is a terrifying warning sign. He's 

also the author of the bestselling book, among many others, from Beirut to Jerusalem. So Tom, 

you write this is potentially as fundamental and significant a turning point as was the emergence 

of mutually assured destruction and the need for nuclear non-prolol proliferation. That's a huge 

statement. How do you say that? And what is your biggest concern? Well, you know, Anderson, 

our um economy now, all our biggest systems, our water systems, our airlines, our airports, 

our transportation and telecommunication systems, all run on software and operating systems as 

do those of uh every other major economy in the world that we are interlin with. And um uh if 

basically we now have a a software that not only is fantastically good at writing new code, but 

it turns out is uh fantastically good at finding bugs in uh in your own code or other people's 

codes. Um that tool, that power now can be used all over the world. Um it would be in the hands 

of everyone. Imagine a world where everyone had a nuclear bazooka basically. Uh and because it's 

so easy, uh a threat, a cyber threat that used to be confined to intelligence organizations, 

big companies was very hard and very expensive to do is become really cheap and really easy. And 

that's why Anthropic Yeah. And anthropic what they they've actually said what they've disclosed is 

that quote the they say the vulnerabilities this program it has spotted have in some cases survived 

decades of human review and millions of automated security tests. I mean that's mind-blowing 

that all the current you know human re humans looking at it and all the the current you know 

AI security tests that they have run programs through and software through all of it still 

had bugs that were be were detected by this new powerful AI well you know this is a point that 

my you know technology tutor and partner in this column Craig Mundy keeps making that this these 

AI capabilities they're coming so much faster faster than people realized and that we aren't, 

as Craig says, birthing just a new tool, Anderson. We're we're birthing a new species. Um it's not 

carbon based like we are, but siliconbased. Um but it is a new species that we're going to 

have to learn to control and collaborate um before it makes us its pet. So this is the front 

end of a really big problem. Well, regulation, control is not a a word that this White House 

is very interested in in terms of artificial intelligence. Uh, anthropic is probably one of the 

the few AI companies, Dario Amade, who left Open AI years ago with his his uh his sister and and 

several others to form Anthropic. They're probably the most kind of safety forward. Um do you think 

that there will be given now this development and others some sort of move to you know really look 

at kind of regulation or or even even government to government? I mean even with some of America's 

adversaries you've written about the idea of the US and China kind of getting together on 

this. Well you know I think Anderson this is the front end of a fundamentally new problem 

we have as a species. Okay. So I started out in journalism and you started out when the world was 

connected. During our tenure with the internet, the world became interconnected. What's happened 

today is the world has become interdependent. Okay? Uh as my teacher dove Sidman likes to say, 

interdependence is no longer our choice. It's our condition. Now um we are going to rise together or 

we're going to fall together. But baby, whatever we're doing, we're doing it together. Now, we can 

discover that early or we can discover that late, but we will discover that. It's true about 

climate, but it's also true now about cyber tech um uh and all of these communication systems. No 

company alone can solve this problem. Therefore, no country alone can solve this problem. 

And therefore, sooner or later, the two tech superpowers, the US and China, are going to have 

to sit down and work out this problem together, learn how to compete and cooperate on AI because 

we are interdependent. That is our condition. It's not a choice. Trump can learn that early. He 

can learn that late, but he will learn that. If you have though these powerful AI weapons at 

your disposal and if it's an American company that has it, if there's an AI arms race, which is 

how many people describe what is happening around the world, isn't there a a kind of a push at some 

point or pressure at some point for an American company to in some way use this for the benefit 

not only of their own company but for you know a country's benefit? The idea of some sort of you 

know China US alliance on this seems far far well difficult to imagine right now. Hard to imagine. 

But um uh you know it's you know when when I say it to people they say god that's incredibly naive. 

I said, "No, no, no, no. What's naive is thinking we're going to be okay if we don't do that because 

there's no way uh both China and America will be hugely vulnerable to criminals within their own 

societies when they have these tools, much more vulnerable than they're going to be to any threat 

the other would pose. So again, we're in the middle of a giant meta transition. You know, I've 

been working a new book on this. We we have become godlike as a species. We we've just created this 

super intelligent being. There's just one problem. We become godlike without the Ten Commandments. 

And you know, we're going to have to sit down and and together with the biggest companies and our 

biggest rivals and write a new Ten Commandments, 10 for us and 10 for the new species we've 

just birthed. OpenAI is calling for governments to implement common sense AI regulation. on 

the substance of the proposal itself. First, what do you make of its approach and how do you 

think lawmakers will respond? Because Congress historically is fairly slowmoving when it comes to 

new technology. Oh, yeah. Slowmoving is one way to put it. At sea is another way to put it, right? 

They have been totally uh incapable of wrapping their arms around what this may represent. I mean, 

you'll remember Boris Andrew Yang who once upon a time ran for president on the concept that we 

would all need to somehow draw from a universal pool of money because AI was going to wipe out 

all jobs. You know, we laughed at him back in the day. That guy must be drinking like a fish these 

days because I mean, it is absolutely the case here that you now have this company saying openly 

that its technology is going to create incredible upheaval. So much so that what it's proposing 

here is pretty interesting, Boris. I mean, it comes up with a number of things. It suggests, 

for instance, that we should all be working a a 4-day work week with no loss of pay, that 

we should uh have a sovereign wealth fund, essentially a kind of general fund that would help 

bring some dividends back to workers who currently aren't seeing anything. And this is a big one 

which breaks from other big tech companies which is the idea that the government needs to stop 

depending on labor payroll as a source of income tax and instead create a a tax entirely on capital 

gains and corporate profits because basically what they're saying here is that you and I and our 

paycheck isn't going to be enough to support the government anymore. All that money is going 

to be vacuumed up into these larger corporations. So there's some good and interesting ideas here. 

as you as you brought up so rightly, you know, the timing here distracting, I think, from a a big 

and embarrassing investigation on the part of the New Yorker is also worth mentioning here. Boris, 

one more question on on the substance of uh the argument for regulation. Alman talks about super 

intelligence, the the point at which AI becomes so sophisticated that it would outco compete or 

outsmart even humans using AI. How close do you think we are to that point? So I am not personally 

convinced that we are anywhere near that. I think that it is some you know the technical people that 

I speak to including people who used to be at open AI and Anthropic and others of these big companies 

basically say we're not on the path to that right now. That the kind of very smart parrot that we 

currently use to write our knock-knock jokes and our wedding vows is not going to put us toward a 

you know an omnisient brain that's going to run all things. But we are already on a path where job 

disruption is real, where human anxiety around AI is real. And so what's interesting here is to 

see a company that is really at the center of that disruption, calling this kind of a new deal 

for AI, trying to invoke the the the Roosevelt era. But of course, that was an era grappling 

with a depression. And this is the company that is essentially saying, "We may cause a depression 

here, and so we're going to need a national response to deal with it." There's a there's a 

a circular logic to this kind of statesmanship on the part of a company that really seems to 

be driving the changes worried about Boris.