US gov suspended access to Fable 5 and Mythos by any foreign national. EU should take note.
Am Wochenende gab es wieder Chaos in der KI-Welt. Die Us-Regierung hat wegen einem narrow (beschränkten, nicht universellen) Jailbreak von Fable 5 und Mythos gefordert, dass Anthropic den Zugang zu diesen KIs für alle Nicht-US-Bürger, sowohl innerhalb als auch außerhalb der USA sperrt.
Da das technisch ohne KYC nicht so einfach differenzierbar ist und auch die EU und andere Verbündete vor den Kopf gestoßen hätte, hat sich Anthropic dafür entschieden, beide Flagship-KI-Modelle vorerst offline zu nehmen.
Was die genauen Hintergründe sind, ob es wirklich ein National-Security-Problem war, ob es von anderen Tech-Unternehmen ausging, oder ob es eine Fortsetzung der Fehde zwischen Dario Amodei und der US-Regierung ist, bleibt umstritten.
Anthropic argumentiert, dass auch andere LLMs von ähnlichen "narrow" Jailbreaks betroffen wären und die US-Direktive überzogen sei.
Jedenfalls ist ein Präzedenzfall, dass die US-Regierung über einen Kill-Switch für KI und IT-Dienstleistungen verfüge.
Was sagt ihr dazu? Was bedeutet das für die EU, wenn die USA den Zugang für KI und andere IT-Dienstleistungen für Nicht-US-Bürger sperren kann?
https://x.com/AnthropicAI/status/2065597531644743999
English
Over the weekend, chaos reigned once again in the AI space. Following a narrow (limited, non-universal) jailbreak of Fable 5 and Mythos, the U.S. government demanded that Anthropic block access to these AI systems for all non-U.S. citizens, both inside and outside the United States.
Since this is technically difficult to differentiate without KYC and would have also offended the EU and other allies, Anthropic decided to take both flagship AI models offline for the time being.
The exact background remains controversial: whether it was truly a national security issue, whether it originated from other tech companies, or whether it is a continuation of the feud between Dario Amodei and the U.S. government.
Anthropic argues that other LLMs would also be affected by similar “narrow” jailbreaks and that the U.S. directive is excessive.
In any case, it sets a precedent that the U.S. government has a kill switch for AI and IT services.
What do you think? What does this mean for the EU if the U.S. can block access to AI and other IT services for non-U.S. citizens at any time?
Fable is extremely pricey. I am sure Anthropic will find their way how to sell it outside the US; they'd lose a bunch of money othervise.
Unless the AI companies find a way to get a billion users to pay the price of a family car each year for AI, most of them are going to Chapter 11. And that's without US government limiting their market.
The U.S. could cut off Mastercard and Visa services to Europe. If they were to suspend these services, any AI outage would pale in comparison.
oh my. more AI on the loose!
Ich hatte wo gelesen, dass es von Amazon ausging, die die Regierung dazu gedrängt hatte.
Europa muss in allen bereichen unabhängiger werden.
They do their interests, and they are right, opposite than EU dumbs
Europe is falling behind.
I don't know, but I don't think they should be blocking access. Then again, governments will do what they want to do.
I think it will have less impact since there are a lot of options still available on AI
In my opinion, this regulation would create a new innovation from non-US citizens which could become the next big thing