Artificial Intelligence (AI) is causing ripples around the globe. From transforming industries to raising ethical dilemmas, AI’s influence is far-reaching. A prominent advocate for AI regulation is Yoshua Bengio, the founder of Mila, the Quebec AI Institute, who recently made an urgent appeal to U.S. lawmakers.
Mr Bengio, an Artificial Intelligence expert, voiced his concerns at a recent hearing in Washington. He urged for an expeditious regulation of AI, highlighting potential risks that could emanate from unbridled technological advancements. His appeal for swift action underscores the gravity of the situation.
Mr Bengio’s plea for regulation comes amid the U.S. Senate subcommittee’s deliberations on potential legislation to control the rapidly evolving AI technology. This technology, while promising unprecedented economic and scientific gains, also poses unimaginable threats.
The spectrum of risks presented by AI is terrifying. It ranges from compromised elections through cyber-attacks to the misuse of nuclear weapons controlled by rogue automated programs. The potential for AI to facilitate large-scale bank fraud and even the creation of biological weapons is also a significant concern.
Bengio’s Policy Recommendations
In light of these fears, Mr Bengio presented a series of possible actions. He argued that international coordination is essential to prevent the misuse of AI technology by rogue actors in countries where the U.S.’s laws do not apply.
Mr Bengio further proposed establishing secure-access international laboratories to investigate countermeasures against the criminal utilization of AI. He insisted that social media companies should be required to authenticate their users, much like banks do with their clients.
Senator Amy Klobuchar, reflecting on the potential for digital identity theft, asked Mr Bengio if he would support legislation giving people control over their image, name, and voice. Mr Bengio agreed but suggested going a step further by establishing strict laws and penalties against counterfeiting humans, akin to the laws against counterfeiting money.
Lastly, Mr Bengio advocated for restrictions on releasing open-source software, which he believes can be manipulated by ill-intentioned individuals for harmful purposes. He cautioned, “If nuclear bombs were software, would you allow open-sourced nuclear bombs?”
The Legislative Landscape
The call for AI regulation has been echoed by other senators, industry executives, and academics. However, passing a bill through the legislative gauntlet of the Republican-led House of Representatives and securing a 60% Senate vote is a daunting challenge.
In the interim, the White House has published voluntary guidelines for AI usage. Last fall, the Biden administration released an AI bill of rights, advocating for the safe and transparent use of AI tools. Additionally, seven major companies, including Amazon, Google, Microsoft, Meta, and OpenAI, have committed to extensive safety testing before releasing a product.
While the U.S. grapples with the complexities of AI regulation, Canada seems to be at the forefront of this legislative race. Bengio informed CBC News that Canada might become the first major AI-producing country with significant AI legislation. Bill C-27, a foundational privacy and data protection law with a section on AI, is currently under consideration.
Several bills proposing various AI regulations are under consideration in the U.S. Congress. These include a bill that would allow Americans to sue AI companies for misusing their likeness in fake videos and a bill that would require companies to share information with researchers. There are also proposals to prohibit the impersonation of election candidates using AI and the use of AI in launching nuclear weapons.