“There is no other solution” for Joe Biden: artificial intelligence “must be supervised”. The White House therefore unveiled rules and principles on Monday intended to ensure that America “leads the way” in regulating this technology.
For the occasion, the president signed a decree which notably requires companies in the sector to transmit to the federal government the results of their security tests, when their projects pose “a serious risk in terms of national security, national economic security, or public health”. The criteria for these safety tests will be set at the federal level and made public.
Identify content produced with AI
“To realize the promises of AI and avoid the risks, we must govern this technology,” declared the leader before signing the decree at the White House, in front of elected officials, members of the government and representatives of industries.
The text also provides guidance on equity (to avoid discriminatory bias from AI), launches research into the impact of artificial intelligence on the labor market and recommends the development of tools to easily identify content produced with AI, in particular.
Biden worries about “deepfakes”
The 80-year-old Democrat also mentioned having seen a video of himself created from scratch with AI (deepfake). “I wondered when could I have said that? “, said Joe Biden, getting emotional about the use of AI to scam people by posing as their family members.
The White House may praise the ambition of the decree, but in reality Joe Biden only has limited room for maneuver. Any truly binding and ambitious legislation on AI should pass through the US Congress. However, the latter is currently divided between Democrats and Republicans, which makes the adoption of a large-scale law very unlikely.
Washington insists on the “moral responsibility” of companies
Since the spring, the White House has insisted on the “moral responsibility” of companies to guarantee the security of their systems. This summer, it also obtained that big names in the digital sector, such as Microsoft and Google, commit to submitting their artificial intelligence systems to external tests.
The decree signed Monday is based on a law dating from the Cold War, the Defense Production Act (1950), which gives the federal government a certain power of constraint on companies, when the security of the country is at stake. Joe Biden further called on parliamentarians to legislate to “protect the privacy” of Americans, at a time when artificial intelligence “not only makes it easier to extract, identify and exploit personal data, but also encourages doing so, since companies use this data to train algorithms.