With the issuance of a new executive order on Monday, US President Joe Biden hopes to mitigate the threats that artificial intelligence (AI) poses to workers, consumers, minority groups and national security.
Joe Biden aims to reduce AI risks with executive order
The order, which Biden signed at the White House, directs agencies to establish standards for such testing and address cybersecurity, radiological, nuclear, biological and chemical risks.
Biden said: “To realize the promise of AI and avoid the risks, we must govern this technology. »
He added: “In the wrong hands, AI can make it easier for hackers to exploit vulnerabilities in the software that runs our society. »
Under the Defense Production Act, developers of AI systems that endanger the economy, public health, safety, or national security of the United States must share security test results with the government before the systems are made available to the general public.
This action represents the administration’s most recent attempt to establish guidelines for artificial intelligence (AI), as the technology rapidly advances in capability and popularity in a so far lightly regulated environment.
Tusk Ventures CEO Bradley Tusk, which invests in technology and artificial intelligence (AI), welcomed the move.
However, he predicted that IT companies would be reluctant to provide the government with confidential information for fear that their competitors would obtain it.
Tusk said: “Without a real enforcement mechanism, which the executive order does not appear to have, the concept is great, but buy-in may be very limited. »
NetChoice, a national trade association of major technology platforms, described the order as an “AI bureaucracy wish list.”
According to NetChoice, this will prevent new companies and competitors from entering the market and significantly expand the federal government’s power over American innovation.
The new order goes beyond voluntary commitments made earlier this year by companies such as OpenAI, Alphabet (GOOGL.O) and Meta Platforms (META.O) to watermark content generated by artificial intelligence (AI) to increase the security of the technology. .
To ensure clarity in government communications, the White House said in a statement that the Commerce Department would “develop guidance for content authentication and watermarking” as part of this order.
The order also requires federal law enforcement agencies and intellectual property regulators to address the use of copyrighted works in AI training, including by calling for “ Evaluating AI Systems to Detect Intellectual Property Law Violations.”
Renowned writers and visual artists have repeatedly sued tech companies, alleging that they stole their designs and used them to train generative AI systems. Tech companies argue that U.S. copyright law’s fair use doctrine protects their use of the content.
The Group of Seven industrialized countries will agree on a code of conduct for companies creating sophisticated AI systems, according to a G7 document.
In response to concerns raised by U.S. officials, Biden’s executive order calls for providing guidance to federal contractors, landlords and federal benefits programs “to prevent AI algorithms from being used to exacerbate discrimination,” according to the press release.
Additionally, the order mandates the creation of “best practices” to address potential harm to workers caused by AI, including job displacement, and requires reporting on labor market impacts.
Subscribe to email updates: