In recent months, Anthropic, an AI research company, has found itself at odds with the U.S. government, sparking significant discussions within the tech community. The conflict began in April when Anthropic unveiled its AI model, Mythos, which it claimed had the potential to create serious cybersecurity challenges due to its advanced coding capabilities. Following a limited release to select cybersecurity experts, the company modified the model and released Fable to the public on June 9. However, just days later, the federal government designated Fable a national security threat, imposing export controls that led Anthropic to revoke access to both models shortly thereafter.
This situation has reignited concerns among critics often referred to as “doomers,” who have long warned about the risks posed by advanced AI technologies. Their fears materialized not through an apocalyptic scenario involving rogue AI but rather through a sophisticated coding model. The rapid government response, which some argue lacks a substantive safety strategy, raises questions about the motivations behind such actions. Notably, Amazon CEO Andy Jassy, whose company has invested in Anthropic and is developing competing AI technologies, played a pivotal role in alerting the government to potential dangers associated with Fable. This situation may not only lead to a temporary ban but could also be challenged legally, as the classification of exporting AI models is not straightforward.
The implications of this feud extend far beyond Anthropic itself. For one, it has caused apprehension among potential users of American AI technologies, prompting some experts to advocate for increased AI development within Europe. However, the ambition to rival Silicon Valley faces challenges, particularly from China’s open-source AI models, which are both powerful and inexpensive. These models can be freely downloaded and deployed, posing a dilemma for companies that may prefer the flexibility of Chinese solutions over the risk of U.S. government intervention. Furthermore, the accessibility of these models could inadvertently increase cybersecurity vulnerabilities rather than mitigate them, as highlighted by leading cybersecurity professionals who argue that Anthropic’s models were essential for strengthening defense strategies.
Finally, the response from U.S. lawmakers will be crucial in the wake of this incident. Following Anthropic’s previous disputes regarding military use of its technology, new legislation has been proposed to clarify the parameters of AI in defense. As the conversation around federal AI regulation gains momentum, public sentiment seems to favor increased oversight. However, the landscape remains uncertain, with varying governmental attitudes towards AI regulation complicating predictions for the future. The changing political climate, illustrated by past shifts in policy from previous administrations, suggests that the upcoming months could lead to further regulatory developments as the government grapples with the complexities of AI safety.
Source: Three things to watch amid Anthropic’s latest feud with the government via MIT Technology Review