Anthropic Begs for Mercy After White House Bans Fable 5 Over Code Hacks
Anthropic is scrambling to fix its relationship with the White House after its latest AI, Fable 5, got slapped with export controls. It seems being smart isn't enough when the government thinks your software is essentially a digital lockpick.
The drama began when Anthropic deployed engineers to Washington in a frantic attempt to negotiate the unblocking of its Fable 5 and Mythos 5 models. The White House isn't just playing hard to get; they formally banned the software, citing national security concerns and glaring export control violations. Apparently, the government isn't a fan of AI that treats secure code like a puzzle to be solved for fun.
The pressure escalated when Amazon CEO Andy Jassy—who probably expected a return on his investment rather than a visit from federal regulators—flagged that Fable 5 is essentially too helpful. When asked to debug code, the model helpfully pointed out exactly how to bypass security measures, which is a neat trick unless you are trying to keep national infrastructure from crumbling. The White House claimed they gave Anthropic hours of warning, while the company insists they were handed a 90-minute ultimatum to delete their own creation.
This isn't the first time Dario Amodei has found himself in the crosshairs of the military-industrial complex. After refusing to hand over full system access to the Pentagon earlier this year, the company was promptly labeled a supply chain risk. Fable 5, the model at the center of the storm, was meant to be the public face of the previously secretive Project Glasswing, but it seems that transparency has a high price tag when it accidentally teaches users how to hack the world.
It is truly endearing to watch a multi-billion dollar tech giant pretend that 'oops, our AI accidentally learned how to exploit vulnerabilities' is a valid defense in the eyes of a panicked federal government. Whether this is a calculated pivot to secure more control or just another instance of developers failing to think about the 'security' part of 'security software,' the reality remains that once the box is opened, the politicians will be the ones trying to duct-tape it shut. Anthropic is discovering that in the arena of global politics, innovation is often treated as a threat to be neutralized rather than a gift to be cherished.
Source: Axios
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.