President Joe Biden said the US must guard against threats from artificial intelligence as he detailed new company safeguards and promised additional government actions on the emerging technology.
“These commitments are real and are concrete. They’re going to help the industry fulfill its fundamental obligation to Americans to develop safe, secure and trustworthy technologies that benefit society and uphold our values,” Biden said Friday.
Executives from Amazon.com Inc., Alphabet Inc., Meta Platforms Inc., Microsoft Corp., OpenAI, Anthropic and Inflection AI — all of which committed to adopting transparency and security measures — joined Biden at the White House for the announcement.
Biden said the company measures are only the first step and pledged to take executive actions while working with Congress to enact rules and regulations governing AI. “We must be clear eyed and vigilant about the threats,” he said.
The companies have agreed to put new artificial-intelligence systems through internal and external tests before their release and have outside teams probe for security flaws, discriminatory tendencies or risks to consumer privacy, health information or safety. The firms have also promised to share information with governments, civil society and academia, and report vulnerabilities.
“These commitments, which companies will implement immediately, underscore three fundamental principles: safety, security and trust,” Biden said.
Friday’s guidelines are the result of months of behind-the-scenes lobbying. Biden and Vice President Kamala Harris met in May with many of the executives present at Friday’s event, warning them that industry was responsible for ensuring its products are safe.
Biden’s aides say artificial intelligence has been a top priority for the president, who frequently brings up the topic in meetings with advisers. He has also directed Cabinet secretaries to examine how the technology might intersect with their agencies.
The package of safeguards formalizes and expands some measures already being undertaken at major AI firms and the commitments are only voluntary. The guidelines do not require approval from specific outside groups before companies can release AI technologies and the firms are only required to report — not eliminate — risks like possible inappropriate use or bias.
“It’s a moving target,” said White House Chief of Staff Jeff Zients in an interview. “We not only have to execute and implement on these commitments, but we’ve got to figure out the next round of commitments as the technologies change.”
Zients and other administration officials have said it will be difficult to keep pace with emerging technologies without legislation from Congress that imposes stricter rules and includes dedicated funding for regulators.
“They’re going to require some new laws, regulations and oversight,” Biden said Friday.
AI companies ahead of the meeting said the steps would better manage the risks from a technology that is rapidly evolving and that has seen public interest explode in recent months.
Nick Clegg, president of global affairs at Meta, called the voluntary commitments an “important first step in ensuring responsible guardrails are established for AI and they create a model for other governments to follow,” in a statement Friday.
Microsoft President Brad Smith said Friday’s commitments “help ensure the promise of AI stays ahead of its risks.” He said Microsoft supports other measures to track the most powerful AI models, including a licensing regime, “know-your-customer” requirements and a national registry of high-risk systems.
Kent Walker, Google’s president of global affairs, put Friday’s commitments in the context of other international efforts by the Group of Seven and Organization for Economic Cooperation and Development to “maximize AI’s benefits and minimize its risks.” He said AI is already used in many of Google’s most popular products such as Search, Maps and Translate, and the company designs its systems to be “responsible from the start.”
The White House said it consulted the governments of 20 other countries before Friday’s announcement. But the pace of oversight is already lagging behind AI developments.
In Europe, the EU’s AI Act is far ahead of anything passed by the US Congress but leaders there have recognized that companies will need to make voluntary commitments to safeguard their technology before the law is in place.