Featured Image Credit: SAUL LOEB / Contributor via Getty
The US government has issued a shocking order that takes down a leading AI model, with the first recall of its kind in history affected Anthropic just days after it launched Claude Fable 5 and Mythos 5.
While AI remains a purely private industry, there has been a strong bond forged between the biggest companies and the US political sphere, with many even supplying services to the government through financially lucrative contracts.
This level of importance does also seemingly have negative consequences, however, as the US government has directly intervened in the release of Anthrophic's latest public AI model, Claude Fable 5, deeming it to pose a risk to national security.
The consequences of this move might prove to be significant – especially considering the fact that Anthropic and the White House are currently locked in an existing legal dispute – and it only continues to emphasize the precarious position that AI currently finds itself within.
Claude Fable 5 is the latest public AI model released by Anthrophic, who are currently the world's most valuable startup company, recently overtaking close rival and ChatGPT creator OpenAI.
Advert
It serves as an extremely limited adaptation of Claude Mythos – a model that remains exclusive to only a select number of companies and government organizations due to its world-leading cybersecurity potential.
Fable has significant guardrails compared to Mythos, and defers to other 'weaker' Anthrophic models for certain prompts or queries, but it is still arguably the most powerful consumer AI model on the market — a position only further cemented by the government's decision to take it down.
As reported by WIRED, the US government revealed that its decision to take down Claude Fable 5 and Mythos 5 was motivated by national security concerns, with the ban initially targeted towards "any foreign national, whether inside or outside the United States, including foreign national Anthropic employees."
This was quickly escalated to all individuals and a complete ban to ensure compliance with the rules, and a new blog post from Anthropic following the ban outlines their interpretation of the government's position."Our understand is that the government believes it has become aware of a method of bypassing, or 'jailbreaking' Fable 5," the statement outlined.
"We reviewed a demonstration of this specific technique being used to identify a small number of previously known, minor vulnerabilities. These vulnerabilities all appear relatively simple, and we have found that other publicly-available models are able to discover them as well without requiring a bypass."
Anthropic continued to emphasize that it worked with the US government previously to "red-team Fable's safeguards for thousands of hours in total," noting that the protections are "substantially more effective than those of any previously deployed model."
The company also noted that there has not been the discovery of any 'universal jailbreak', deeming it not currently possible for any existing model provider.
While the team behind the powerful AI tool is working to restore access as soon as possible, it only continues to reflect the tensions between Anthropic and the White House.
The AI company is in the middle of a lawsuit against the US government after designation banned government agencies and contractors from using its models, with the Department of Defense branding Anthropic a 'supply chain risk' after it outlined restrictions for the use of its tools within the military.
Anthropic demanded that its AI must not be used to both spy on Americans and develop weapons that can be fired without human involvement – both reasonable requests – yet caused a significant rift to emerge that might be a motivating factor behind the Fable 5 ban.
