In a significant move to shape the future of artificial intelligence (AI) in the United States, President Biden has enacted a new executive order aiming to introduce stringent oversight on AI usage and development.
This legislation is particularly focusing on models that have not yet been released, such as the highly anticipated GPT-5 from OpenAI.
The executive order, signed on Monday, sets forth a framework for governing how the federal government employs AI and outlines clear guidelines for companies that are in the process of building new models.
During a White House event, President Biden addressed a gathering that included lawmakers, industry stakeholders and members of the press.
He articulated the intention behind the executive order: "I’m determined to do everything in my power to promote and demand responsible innovation," and highlighted AI as the "most consequential technology of our time".
National security a concern
Under the new regulation, companies developing AI models that could potentially pose a risk to national security are now required to disclose such models to the government.
They must also provide data concerning the security measures being adhered to, in alignment with federal standards that the National Institute of Standards and Technology will develop.
This decree mandates the sharing of pre-release testing data for models that are yet to be launched, directly implicating the forthcoming GPT-5 model.
President Biden stressed the importance of this measure: "Companies must tell the government about the large scale AI systems they’re developing and share rigorous independent test results to prove they pose no national security or safety risk to the American people."
Currently deployed models not exempted
Ben Buchanan, AI special advisor to the White House, in a conversation with Forbes, noted that currently deployed models, such as GPT-4 or Google’s Bard, are not exempt from the executive order.
They are subject to the full extent of the regulations, which include provisions for equity, non-discrimination, and the protection of consumers and workers.
Buchanan added, "we've not seen a Chat GPT-4 enabled catastrophe that I know of," indicating the relative stability of existing AI systems.
Substantial recruitment drive
The order also sets in motion a substantial recruitment drive for AI experts within the federal government, aiming for "dozens to hundreds" of new hires specifically focused on AI.
It also intends to lower the hurdles for AI professionals seeking to immigrate to the United States.
While this does not increase the cap on H1B visas, Buchanan pointed out that there would be a greater emphasis on enhancing the overall visa application experience for individuals working with "critical emerging technologies," potentially bolstering the United States' position in the global AI race.