Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

U.S. allows Anthropic to release Mythos AI to ‘trusted’ US organizations

semafor.com

Post reply on HN
Print view View original

501–510 of 873 posts

Re: U.S. allows Anthropic to release Mythos AI to ‘trusted’ US organizations

#501

Post by RobotToaster »

Maxious wrote: post #498
nazgob wrote: post #384

Earlier quoted context omitted.

Does it apply to USA citizens and organizations as well?

Fun fact the 1996 cryptography restrictions have never been repealed. You just haven't heard of them as open source software has broadly been exempted https://www.linuxfoundation.org/resources/publications/under...

I once had to go through the process of buying a compiled encryption library from a US company, which involved filling in a multi page form basically pinky swearing I wasn't a terrorist and wasn't going to sell it to North Korea.

On the bright side I got to add "successfully managed the purchase of controlled munitions" to my CV.

Re: U.S. allows Anthropic to release Mythos AI to ‘trusted’ US organizations

#502

Post by dgellow »

notrealyme123 wrote: post #495
re-thc wrote: post #486

Earlier quoted context omitted.

Doesn’t help unless they get GPUs or equivalent to train.

With enough money this problem is solvable. GPUs alone are only a fraction of a frontier model.

Im really not convinced that can be solved with money alone. China has the money and infrastructure, they don’t have the chips. Europe doesn’t have the infrastructure or the money but may have some chips

Re: U.S. allows Anthropic to release Mythos AI to ‘trusted’ US organizations

#503

Post by dgellow »

OtherShrezzing wrote: post #426

What’ll happen when Google Deep Mind go to release their next models, developed mostly in London? Are Google going to end up in a situation where the people working on their models cannot use the models after launch?

It’s whatever the Trump admin decides. There are no rules here, we are past that

Re: U.S. allows Anthropic to release Mythos AI to ‘trusted’ US organizations

#504

Post by dgellow »

varjag wrote: post #395
sschueller wrote: post #382

Earlier quoted context omitted.

The LLM improvement curve will level off and with so many AI engineering being in China they will catch up quickly. The financial growth based on that everyone on earth will need to have an OpenAI or Anthropic subscription is quickly falling apart. DeepSeek V4 is good enough for most of my work that that I can no longer justify paying 100x for something else. The cost difference is astronomical.

Not sure why people here are so upbeat about China. It is almost guaranteed it will also introduce export controls on the models.

Right now they are the one competing openly

Re: U.S. allows Anthropic to release Mythos AI to ‘trusted’ US organizations

#505

Post by roenxi »

moooo99 wrote: post #414
varjag wrote: post #395

Earlier quoted context omitted.

Not sure why people here are so upbeat about China. It is almost guaranteed it will also introduce export controls on the models.

> It is almost guaranteed it will also introduce export controls on the models. The current models are open weights and already out the door. They are hosted by many providers and are already comparatively good in many domains. Even if this generation is the last one to be open, I‘d argue it would already put the US providers in trouble

Also, the evidence isn't that China is doing anything magic - it seems AI is just energy + compute as opposed to any special research edge. There is reason to think that anyone would be capable of building these models. They're generic and every country will eventually catch up at a speed depending on how economically capable they are at buildign data centres.

I'm upbreat about China because they seem to be the biggest player here, but even if they don't come through I expect other countries will be able to put out decent free models.

Re: U.S. allows Anthropic to release Mythos AI to ‘trusted’ US organizations

#506

Post by b112 »

qtk8 wrote: post #481
0000000000100 wrote: post #461

Earlier quoted context omitted.

Marketing stunt by who dude? Do you really think Anthropic would really go this far? This whole situation is completely absurd. The federal government is arbitrarily restricting AI models without really providing any reasoning. This is not a clear and transparent government and it just feels like gears are moving behind the scenes that have lead this out of character restriction of private and extremely wealthy (at l…

It could be that the us govt doesn't want china to execute distillation attacks and narrow the gap.

Next gen models are greatly aided by data obtained from existing model use.

By restricting use, the outcome may be slower progress, thus narrowing the gap for other reasons.

Re: U.S. allows Anthropic to release Mythos AI to ‘trusted’ US organizations

#507

Post by win311fwg »

varjag wrote: post #395
sschueller wrote: post #382

Earlier quoted context omitted.

The LLM improvement curve will level off and with so many AI engineering being in China they will catch up quickly. The financial growth based on that everyone on earth will need to have an OpenAI or Anthropic subscription is quickly falling apart. DeepSeek V4 is good enough for most of my work that that I can no longer justify paying 100x for something else. The cost difference is astronomical.

Not sure why people here are so upbeat about China. It is almost guaranteed it will also introduce export controls on the models.

Export controls are only a strategic advantage if the export destinations don't already have the same (or better) technology. That time may come, but first China needs to export to destroy the AI industry in the USA.

Re: U.S. allows Anthropic to release Mythos AI to ‘trusted’ US organizations

#508

Post by cookiengineer »

qtk8 wrote: post #482
DSingularity wrote: post #429

Earlier quoted context omitted.

[flagged]

They got a democratic mandate.

> They got a democratic mandate.

Fun fact that is missing in US history education: Hitler was a legally elected chancellor.

Don't skip history class, kids.

Re: U.S. allows Anthropic to release Mythos AI to ‘trusted’ US organizations

#509

Post by lstodd »

watwut wrote: post #493
ben_w wrote: post #479

Earlier quoted context omitted.

While I also regard the "doom from the companies themselves is just a marketing stunt" arguments as conspiracy-theory territory (especially since neither OpenAI nor Anthropic* changed their tune since before they were rich enough for meaningful lobbying): This particular specific doom is from the USG, Trump has a history of kayfabe, and there's a stink of market manipulation coming from the White House. * Musk, howev…

So, assuming they are moderately sane is "conspiracy theory"? Assuming they are evil unsane and actually believe that crap while doing what they do looks way more like conspiracy theory.

Do not assume people are sane for any definition of sane. That way lies the madness.

Re: U.S. allows Anthropic to release Mythos AI to ‘trusted’ US organizations

#510

Post by klohto »

baq wrote: post #466

I’m on record saying mistral needs 50B euros asap; I’ll admit this was wrong, they need more sooner

To do what exactly? They have been nothing but EU funds leech and corporiders.

Every single model is (and always was) behind anything from China.

I wish Europe could find one company that can put us on the map but Mistral’s not it. They have repeatedly shown that extra funding doesn’t magically transfer to better models.

Post reply on HN