00:00Trump just lifted the ban on Claude Fable 5 and it's officially back,
00:04so Anthropik's most powerful public model goes live again, but it's still not fully unleashed.
00:09Their own post mentions that some routine tasks, such as coding and debugging, will fall back to
00:14Opus 4.8. Anthropik's cloud code lead tried to clarify it but made it vaguer by saying a small
00:20fraction of coding and debugging tasks will get flagged and routed to Opus instead. So, which
00:25tasks and why? Those are not clarified. Also, getting Fable back came with conditions and here
00:32are the three things Anthropik agreed to. First, they have to proactively detect and address security
00:37risks tied to Fable 5 and Mythos 5. However, how they're actually required to do that hasn't been
00:43made public. Second, they agreed to work with the US government on the protocols, standards,
00:48and releases for Mythos, Fable, and future models. So it seems to cover basically everything Anthropik
00:55ships next. Finally, they must report any malicious activity involving the models to the government.
01:00Here's why that matters. A few weeks ago, the US government paused a private company's AI launch.
01:06And now that pause is embedded into how that company's future models are released.
01:10That's the government sitting inside the release process of a frontier AI lab.
01:14We haven't really seen that before. And it opens a bigger question. If this holds for Anthropik,
01:19what about every other groundbreaking AI that ships in the US? Does the government get a say in those too?
01:25Comment your thoughts, and for more AI updates, subscribe and follow Alpha Intellect.
Comments