Two US Senators, Richard Blumenthal and Josh Hawley, have introduced a legislative framework aimed at regulating the development and implementation of artificial intelligence (AI) technology. The lawmakers propose that developers of AI applications, particularly face recognition technology and other “high-risk” AI systems, should be required to obtain a government license. This license would necessitate harmness testing of AI models, independent auditing of products, transparency regarding training data, and the provision of legal recourse for individuals harmed by AI systems.
The proposed framework has garnered support from AI ethics expert Anna Lenhart, who commends the Senators for taking proactive steps rather than relying on lengthy and potentially ineffective consultation processes. However, Lenhart also acknowledges the challenges associated with having a single legislative body equipped with the necessary technical and legal expertise to oversee licensing for a wide range of AI applications.
Blumenthal and Hawley are also set to oversee a Senate subcommittee hearing dedicated to discussing accountability for AI technology that causes harm or infringes on users’ rights. The hearing will explore ways to hold both businesses and governments accountable for the responsible implementation of AI systems. Additionally, Senator Chuck Schumer has organized a series of discussions focused on AI regulation, inviting technology leaders such as Mark Zuckerberg and Elon Musk, as well as CEOs from companies like Google and Microsoft, researchers, and human rights experts to participate. These events indicate a growing interest in regulating the potentially significant powers of AI.
The current leadership structure of the US Cyber Command (Cybercom) and the National Security Agency (NSA) is also under scrutiny. Since its establishment in 2010, Cybercom has been led by the same individual who heads the NSA. Initially considered a temporary solution to leverage the existing resources and personnel of the NSA, this dual-hat arrangement has remained in place for over a decade. Security experts have weighed the pros and cons of this leadership structure, with contrasting opinions emerging.
Proponents argue that the shared leadership arrangement enables more efficient decision-making, particularly in the face of swiftly evolving cyber threats. Senator Mike Rounds highlights the potential clash between separate bureaucracies if the dual-hat arrangement were dissolved, emphasizing the importance of coordination and protection in a cyber landscape. Army General Paul Nakasone, who has held positions at both Cybercom and NSA, champions the arrangement, claiming that without it, the cost and effectiveness of protecting national security in cyberspace would be compromised.
On the other hand, critics argue that placing one individual in charge of both institutions concentrates too much power in one person, especially as Cybercom gains more authority. The concern is that Cybercom’s connection to the NSA could expose secretive intelligence activities to adversaries. Additionally, there is a fundamental tension between an intelligence organization like the NSA, focused on collecting and analyzing information, and an operational organization like Cybercom, with a mission to effect significant action.
Despite the ongoing debate, significant changes to the leadership structure are unlikely in the short-term. Congress has established specific metrics that must be achieved before a split can occur, and these metrics have yet to be met. Furthermore, General Nakasone’s plan to step down this year and the current Senate stalemate on military appointments further complicates the situation.
In conclusion, the introduction of a legislative framework by Senators Blumenthal and Hawley demonstrates a growing recognition of the need to regulate AI technology. This proposal, if enacted, would require developers of high-risk AI applications to obtain a government license, undergo harmness testing, independent auditing, and provide transparency. Additionally, ongoing discussions led by Senator Schumer and the upcoming Senate subcommittee hearing indicate a broader interest in addressing the challenges associated with AI regulation.
The debate surrounding the shared leadership structure between Cybercom and the NSA highlights conflicting opinions on efficiency and concentration of power. Proponents argue that the arrangement facilitates coordination and protection, while critics raise concerns about intelligence exposure and the inherent tension between intelligence and operational activities. Nonetheless, substantial changes to the current arrangement are unlikely in the near future, as specific metrics need to be met, and military appointments are currently at a standstill.