OpenAI on Monday released what it calls an “economic blueprint” for AI, a living document outlining the policies it believes it can build in partnership with the U.S. government and its allies.
The blueprint also includes remarks from Chris Lehane, OpenAI's vice president of international affairs, who says the U.S. will spend billions of dollars on chips, data, energy and talent needed to “win with AI.” They argue that action must be taken to attract funding for the future.
“Today, while some countries are ignoring AI and its economic potential, the U.S. government can protect national security while paving the way for the AI industry to continue its global leadership in innovation. We can,” Lehane wrote.
OpenAI has repeatedly called on the U.S. government to take more substantive action regarding the infrastructure that supports AI and technology development. The federal government has largely left AI regulation to the states, and OpenAI explains in its blueprint that this situation cannot be maintained.
In 2024 alone, state legislatures introduced nearly 700 AI-related bills, some of which conflict with other bills. For example, Texas' Responsible AI Governance Act imposes onerous liability requirements on developers of open source AI models.
OpenAI CEO Sam Altman has also pushed for existing federal laws, such as the CHIPS Act, which aims to revitalize the U.S. semiconductor industry by attracting domestic investment from the world's top chip makers. criticizing. In a recent interview with Bloomberg, Altman said of the CHIPS Act:[has not] “It was as effective as any of us could have hoped,” he said, adding that he believed the Trump administration had a “real opportunity” to “do better as a successor.”
“What I really and deeply agree with is that [Trump] And what's more, it's how difficult it's become to make things in the United States,” Altman said in an interview. “Power plants, data centers, things like that. I understand how bureaucratic corruption accumulates, but it doesn't help the country as a whole. What needs to happen for the U.S. to lead in AI? It's not particularly helpful when you think about whether there is a need for the US to really lead in AI.
To fuel the data centers needed to develop and run AI, the OpenAI blueprint would “significantly” increase federal spending on electricity and data transmission, and increase funding for things like solar power, wind farms, and nuclear power. It recommends meaningfully building “new energy sources.” OpenAI, along with its AI rivals, has previously supported nuclear power projects and argued that nuclear power is needed to meet the power needs of next-generation server farms.
Tech giants Meta and AWS have stalled on their nuclear efforts for reasons unrelated to nuclear power itself.
In the near term, OpenAI's blueprint will allow governments to “develop best practices” to prevent model abuse, “streamline” the AI industry's engagement with national security agencies, and allow models to be shared with allies. It is proposed that export controls be developed to limit[ing]Export to “hostile countries”. Additionally, the blueprint will help governments share certain national security-related information with vendors, such as a description of threats to the AI industry, to ensure vendors have the resources to assess the risk of their models. is encouraged.
“The Federal Government’s approach to safety and security in the Frontier Model should streamline requirements,” the blueprint reads. “Responsibly exporting models to our allies and partners will enable us to innovate with AI and share its benefits, while building AI on American technology rather than technology funded by the Chinese Communist Party.” It will help you launch your own AI ecosystem, including your homegrown developer community.”
OpenAI already counts several U.S. government departments as partners and plans to add more as its blueprint gains traction among policymakers. The company has contracts with the Department of Defense for cybersecurity work and other related projects, and has partnered with defense startup Anduril to bring its AI technology to systems used by the U.S. military to counter drone attacks. are.
In its blueprint, OpenAI calls for drafting standards on behalf of the U.S. private sector that will be “recognized and respected” by other countries and international organizations. However, the company stopped short of endorsing mandatory rules or edicts. “[The government can create] Defined voluntary pathway for companies conducting development [AI] “We will work with governments to define model assessments, test models, and exchange information to support enterprise safety measures,” the blueprint reads.
The Biden administration took a similar path with its AI Executive Order, calling for the establishment of several high-level voluntary AI safety and security standards. The executive order established the US AI Safety Institute (AISI), a federal agency that studies risks in AI systems and partnered with companies such as OpenAI to assess the safety of models. But President Trump and his allies have vowed to repeal the Biden executive order, putting its codification and AISI at risk of being scrapped.
The OpenAI Blueprint also addresses the hot topic of copyright related to AI. The company argues that AI developers should be able to use “public information,” including copyrighted content, to develop models.
OpenAI, like many other AI companies, trains its models based on public data from across the web. The company has licensing agreements with many platforms and publishers that provide limited ways for creators to “opt out” of model development. However, OpenAI also says it is “impossible” to train AI models without using copyrighted material, and many creators have accused the company of training their work without permission. I'm appealing.
“[O]“Parties, including developers from other countries, have made no effort to respect or engage with intellectual property owners,” the blueprint reads. “If the United States and like-minded countries do not address this imbalance through smart steps that support the long-term advancement of AI, the same content will be used for AI training in other regions. will be used for other economic benefits. [The government should ensure] Like humans, AI has the ability to learn from universally available information, while also protecting creators from unauthorized digital replicas. ”
It remains to be seen which parts of OpenAI's blueprint will influence legislation. However, these proposals indicate that OpenAI intends to remain an important player in the race for a unified U.S. AI policy.
OpenAI more than tripled its lobbying spending in the first half of last year, spending $800,000 compared to $260,000 in all of 2023. The company also added former government leaders to its ranks, including former Pentagon official Sasha Baker and NSA Director Paul Paul. Nakasone and Aaron Chatterjee, who served as chief economist at the Commerce Department under former President Joe Biden.
As OpenAI hires talent and expands its international affairs department, it has become more vocal about which AI laws and regulations it prioritizes, including establishing a federal rulemaking agency for AI and He supports a Senate bill that would provide federal scholarships to students. Research and development. The company has also opposed legislation, particularly California's SB1047, saying it would stifle AI innovation and drive out talent.