The White House said it had obtained voluntary agreements from seven leading companies that operate there artificial intelligence (AI) promote greater security, transparency and trust in the development of new technology.
Key things
- Executives from Amazon, Anthropic, Alphabet, Inflection, Meta, Microsoft and OpenAI will meet with administration officials at the White House on Friday.
- Companies have agreed to several AI safeguards, such as watermarking and external testing.
- The White House also said it is preparing an executive order on AI and will work on bipartisan legislation to support innovation.
Amazon management (AMZN), anthropic, alphabet (GOOG, GOOGLE), inflection, meta (META), Microsoft (MSFT), a OpenAI will meet with administration officials at the White House on Friday to discuss the deal, which covers a wide range of potential challenges posed by AI, from verifying content to ensuring the safety and security of new models.
One of the main terms of the agreement is that the companies commit to internal and external reviews of their AI systems before release. While the agreement is voluntary, a White House official said the external verification is a way for officials to know whether the companies are following their commitments.
“This is the next step we’re taking, but it’s not the last step,” a White House official said.
The White House agreement requires companies to invest in both cybersecurity and internal safeguards while committing to third-party vulnerability testing. Companies are also committed to sharing information across industry and with government, academia and organizations that will manage the risks AI poses to society.
Another key element of the White House deal is making sure the public knows when content is generated by AI, such as the watermarking system.
While a White House official said companies already have AI security standards in place, the deal will add new requirements for some companies, such as with watermarking.
“It’s fair to say that this is pushing the boundaries of what companies are doing and raising the bar for safety, security and trust in AI,” a White House official said.
In addition, the White House said it is working on an executive order to help give the federal government more authority to manage AI risks and is advancing bipartisan legislation designed to encourage responsible innovation in the technology.
The agreement is the latest development in the White House’s push to advance artificial intelligence regulation. For months, the administration has been meeting with tech companies about AI development, including a meeting with Vice President Kamala Harris May at the White House.