Apple signs the White House's commitment to AI security

Apple recently joined the White House by signing a voluntary commitment to develop safe, reliable, and trustworthy artificial intelligence. According to the press release, the company plans to integrate its generative AI technologies, called Apple Intelligence, into its core products, providing them to two billion users.

Apple is among the 15 technology companies, including Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and OpenAI, that agreed to adhere to the White House's ground rules for developing generative AI in July 2023. At that time, Apple had not disclosed the extent of its AI plans for iOS. However, at the WWDC conference in June, the company made it clear that it is serious about integrating generative AI, starting with a partnership that will embed ChatGPT in the iPhone. As a frequent target of federal regulators, Apple wants to show its willingness to follow the White House rules in advance, possibly in an attempt to avoid future regulatory disputes over AI.

But how significant are Apple's voluntary commitments to the White House? They do not have much force yet, but it is a start. The White House calls this the "first step" towards developing safe, reliable, and trustworthy AI. The second step was President Biden's AI executive order in October, and several bills are currently being considered in federal and state legislatures to better regulate AI models.

As part of the commitment, AI companies promise to red-team (acting as adversarial hackers to stress test an organization’s safety measures) AI models before public release and share that information with the public. The White House’s voluntary commitment also requires companies to handle unreleased AI model weights confidentially. Apple and other companies agree to work on AI model weights in secure environments, limiting access to them to as few employees as possible. Finally, the companies agree to develop content labeling systems, such as watermarking, to help users distinguish what is AI-generated content from genuine content.

Separately, the U.S. Department of Commerce will soon release a report on the potential benefits, risks, and implications of open-source AI models. Open-source AI is increasingly becoming a politically charged regulatory arena. Some groups want to limit the accessibility of weights of powerful AI models in the name of safety. However, this could significantly limit the startup and research ecosystem in the AI field. The White House's stance on this issue could have a significant impact on the entire AI industry.

The White House also noted that federal agencies have made significant progress in meeting the tasks set out in the October executive order. Federal agencies have hired more than 200 AI specialists, provided more than 80 research teams with access to computational resources, and released several frameworks for developing AI.