Donald Trump said that in the heated debate over the use of intellectual property in its training models, AI companies cannot expect to pay for the use of copyrighted content in their systems.
Speaking at AI Summit on Wednesday, Trump said: “If every article, book or everything you study is expected to pay, you can't expect an AI program to succeed. But you can't do that, so you can't.
Trump called for what he characterized as “common-sense artificial and intellectual property rules,” saying the US can't afford to fall behind China, which doesn't have strong IP protections.
“When people read books and articles, you gain a lot of knowledge. That doesn't mean you have to violate copyright laws or have to deal with all content providers,” he said. “You can't do that. China isn't doing that.”
His speech was to mark the White House release of the AI Action Plan, which prioritizes building the country's AI capabilities, including data centers and other support, while removing regulatory barriers.
Related: Senators introduce bills to limit the misuse of copyrighted works for AI companies' training models
The plan is in contrast to Trump's predecessor, Joe Biden. Joe Biden focuses on the government's role in ensuring technology is safe.
The Trump White House plan recommends that the government update federal procurement guidelines to ensure that governments only contract with frontier leading language model (LLM) developers. It also recommends revising the National Standards and Technology AI Risk Management Framework to remove misinformation, DEIs and references to climate change.
“We must ensure that freedom of speech flourishes in the age of AI and that the AI sourced by the federal government objectively reflects the truth, not the agenda of social engineering,” the plan says.
However, what is not mentioned in the plan is copyright, and the concern among content creators is how the material is used in the training model. That led to numerous lawsuits against authors and studios such as Disney and NBCUniversal against AI companies.
Related: Amazon invests another $100 million in AWS Generic AI Innovation Center
On Monday, Sens. Josh Hawley (R-MO) and Sen. Richard Blumenthal (D-CT) announced laws requiring AI companies to obtain personal consent before using content and data in the development of AI systems.
But Trump's remarks reflect opposition to such an approach. Openai has signed licensing agreements with companies like News Corp. The Associated Press content, but the technology industry argues that using content in its training model is “fair use.”
Trump said, “When you have something and read something, and when it enters this vast intelligence machine, when we call it, you say, “Let's pay this so much,” you can't expect every time. Of course, it doesn't work that way. Articles cannot be copied or plagiarized, but they need to enable AI to use that pool of knowledge without experiencing the complexity of contract negotiations.
The AI Action Plan “refers to the risk of deepfakes, whether audio recordings, photo videos, and the plan cites the Take It Down Act, which was recently handed over, aimed at reducing sexually explicit and nonconsensual deepfakes, while the AI Plan cites the need for “additional actions.” Specifically, it is to establish formal guidelines and a “voluntary forensic framework” for use in determining the truthfulness of evidence in legal proceedings.
The plan doesn't focus on the safety that Biden approach has done, but the White House AI proposal outlines risks in areas such as national security and personal privacy.
“The United States must lead the creation of the world's largest and highest quality AI Ready science dataset, while maintaining respect for individual rights and ensuring civil liberties, privacy and security,” the plan states.
The plan also aims to regulate states, saying the federal government should not “direct to the state with AI-related federal funds with burdensome AI regulations that waste these funds, but should not interfere with the right to pass prudent laws that are not overly restricted to innovation.” FCC Review State Law and the FTC recommend that final orders, agreements and injunctions be “unfairly borne AI innovation.”
In his speech, Trump said the United States must have a single federal standard, not 50 different states regulating this future industry.
California Governor Gavin Newsom has fired Trump about the attack on state laws of the AI Program, saying, “It threatens to bring back states like California with powerful laws against child pornography that have been generated, and might say it's an interesting priority, especially in light of his close relationship with Jeffrey Epstein.”
