This analysis is based on breaking news and will be updated. To connect with the author, please e-mail [email protected].

The Biden administration today released a long-awaited major Executive Order (EO) on “Safe, Secure, and Trustworthy Artificial Intelligence (AI),” the latest effort by the White House to unilaterally advance AI policy as Congress continues to struggle with the issue. Despite intense interest and a flurry of proposals, comprehensive AI legislation appears unlikely in the near term. The White House has been looking to fill the federal AI policy vacuum through various policy statements, agency actions and other steps. Biden’s new 100-plus page EO includes “sprawling directives to over a dozen agencies,” ordering them to look into implications of algorithmic systems and processes for a wide variety of issues including copyright, competitiveness, cybersecurity, education, health, housing, infrastructure, labor and privacy.

While some will appreciate the order’s whole-of-government approach to AI, unilateral and heavy-handed administrative meddling in AI markets could undermine America’s global competitiveness—and even the nation’s geopolitical security—if taken too far. AI is a critical new technology with the potential to expand productivity and economic growth fundamentally, with benefits accruing across many sectors and for all consumers. AI has particularly important implications for advancing public health. AI and computational science also have national security ramifications, which is why a strong and secure domestic technology base is essential to countering challenges or threats from China and other nations. Excessive preemptive regulation of AI systems could impede the growth of these technologies or limit their potential in various ways.

Dystopian Narratives Driving Calls for AI Regulation

Unfortunately, the policy debate around AI thus far has been mostly driven by worst-case scenarios pulled from the plots of dystopian science fiction books and movies. These fears have triggered calls for sweeping regulatory controls for AI—including new regulatory agencies, licensing schemes and expanded liability—that will hamper the adoption of beneficial new algorithmic applications significantly.

The EO comes just two days before the United Kingdom’s AI Safety Summit, a multi-nation effort focused on addressing risks with more powerful “frontier AI” technologies, or advanced supercomputing systems. Wired refers to the coming summit as “a doom-obsessed mess” due to its focus on so-called “existential risks” and extreme regulatory steps to address them. Surprisingly, the U.K. government has transformed quickly from a leading exponent of light-touch regulation for AI systems to a platform for some of the most extreme solutions for controlling computation—part of a new effort under Prime Minister Rishi Sunak to “write the AI rulebook” for the world. The Biden administration has been coordinating with the U.K. government on this and other AI issues, but whether it steers the United States down a similarly radical regulatory path remains to be seen.

Thus far, the Biden administration has focused primarily on pressuring major AI innovators to make voluntary concessions regarding AI safety and data sharing. Building on this, the new EO stretches the broad stipulations of the Defense Production Act to require that “companies developing any foundation model that poses a serious risk to national security, national economic security, or national public health and safety must notify the federal government when training the model, and must share the results of all red-team safety tests.” Developers already use red-team safety tests regularly to stress-test AI models for deficiencies and make corrections. The White House is now pushing to formalize this process through standards created by the National Institute of Standards and Technology (NIST) and applied by agencies like the Department of Energy and the Department of Homeland Security through an AI Safety and Security Board. The EO also contains vague language about the need to “accelerate development and implementation of vital AI standards with international partners” in an effort to “establish robust international frameworks” for AI, but it offers few details before simply noting that Vice President Kamala Harris will participate in Wednesday’s AI Safety Summit.

The EO focuses on other priorities beyond just frontier AI systems, however. First, it calls for broad-based efforts to expand privacy protections, including through the development of additional guidelines on how algorithmic systems collect and use data. Other provisions in the EO call for enhanced cybersecurity protections to expand upon the “AI Cyber Challenge” program launched this summer to “challenge competitors across the United States to identify and fix software vulnerabilities using AI.” The EO also includes requirements that the military and intelligence community “use AI safely, ethically, and effectively in their missions.” Finally, it calls for greater efforts to address fraudulent or deceptive uses of AI and calls on the Department of Commerce to craft guidance for content authentication and watermarking to clearly label AI-generated content.

On other issues, the new EO maintains the administration’s rhetorical approach—initially sketched out in its October 2022 “Blueprint for an AI Bill of Rights,” which focused on fears about AI and claimed algorithmic systems are “unsafe, ineffective, or biased” and “deeply harmful.” For example, fearing that AI systems will “exacerbate discrimination” in housing, the workplace or federal benefits programs, the EO calls for additional steps to combat “algorithmic discrimination.”

Empowering the Administrative State

The new EO highlights the administration’s adoption of an everything-and-the-kitchen-sink approach to AI policy that is extremely ambitious as well as potentially overzealous. The implementation details are mostly left to the various federal agencies to work out, and it remains unclear how far they can stretch their statutory authority to enforce many of these stipulations.

Even so, taken together with other recent administration statements, the EO represents a potential sea change in the nation’s approach to digital technology markets, as federal policymakers appear ready to shun the open innovation model that made American firms global leaders in almost every computing and digital technology sector. With the United States now facing fierce competition from global AI companies in China and other nations, the danger exists that the country could put algorithmic innovators in a regulatory cage, encumbering them with many layers of bureaucratic permission slips before any new product or service could launch. Biden’s new EO could accelerate the move to tie the hands of algorithmic entrepreneurs even if Congress does not pass any new legislation on this front.

There are some positive and much-needed elements to the EO, however, including its call “to expand the ability of highly skilled immigrants and nonimmigrants with expertise in critical areas to study, stay, and work in the United States by modernizing and streamlining visa criteria, interviews, and reviews.” For some time, there has been a pressing need to expand efforts to retain skilled immigrant workers, with many technology companies and experts worried about losing top-notch talent to other nations.

But most of the EO focuses on broader and extremely vague calls for expanded government oversight across many other issues and agencies, raising the risk of a “death by a thousand cuts” scenario for AI policy in the United States. For example, while there is nothing wrong with federal agencies being encouraged through the EO to use NIST’s AI Risk Management Framework to help guide sensible AI governance standards, it is crucial to recall that the framework is voluntary and meant to be highly flexible and iterative—not an open-ended mandate for widespread algorithmic regulation. The Biden EO appears to empower agencies to gradually convert that voluntary guidance and other amorphous guidelines into a sort of back-door regulatory regime (a process made easier by the lack of congressional action on AI issues).

Of greater concern is the EO’s green light for the Federal Trade Commission (FTC) to expand its focus on AI policy. While the agency does possess broad powers to police unfair and deceptive practices within all markets, the administration’s call for it to exercise greater regulatory authority over the AI ecosystem creates the potential for preemptive overreach. The FTC’s controversial Chair, Lina Khan, has radicalized the agency and pursued aggressive actions against digital technology companies since her tenure began. The FTC has made it clear that AI systems are in its sights, and the agency could be positioning itself to serve as America’s de facto AI regulator. Because the Biden administration’s new EO (in addition to its previous AI Bill of Rights) suggests that broad-based harms are omnipresent within algorithmic systems, it could serve as an open-ended invitation for the FTC to overzealously harass AI innovators and micromanage developing markets.

Meanwhile, the new EO hints at how agencies could use federal procurement procedures as an indirect method of AI regulation. Because the federal government invests significant resources in digital systems through grants and contracts, it gives the Executive Branch considerable leverage to dictate how that money is used by private parties—including how they develop AI. Analysts have pointed to the inherent risks of politicizing procurement policies and using the so-called “power of the purse” to shape social or market outcomes, however. Worse yet, rigging procurement rules to steer technology decisions to achieve predetermined political preferences or market outcomes could undermine the benefits associated with the rapid development and diffusion of algorithmic technologies.

Shooting Ourselves in the Foot as the Race Gets Underway

The Biden administration’s actions this week, from releasing its new EO to negotiating with the U.K. at the AI Safety Summit, could go down as a historic moment in the history of technology policy—but perhaps not in a positive way. Many nations taking part in Wednesday’s AI summit are looking to aggressively regulate leading algorithmic technologies and innovators, many of which are U.S.-based. “Every other country that will be represented at the summit wishes it had a technology industry like we have in the United States,” argues Wall Street Journal columnist James Freeman. “This means we have more to lose by far than anyone else if the direction of technological development is moved from the marketplace to the halls of governments. This ought to give U.S. politicians pause before joining such multinational efforts.”

This is particularly true of the European regulators looking to aggressively regulate U.S. tech companies in recent years. The European Union (EU) has become an innovation backwater, with almost no leading digital innovators to show for all their regulatory and industrial policy efforts. The continent’s leading digital export is now regulation, not world-class products or services—forcing some analysts to conclude that the EU has become “The Biggest Loser” in the global digital technology race, stating that “the future will not be invented in Europe.” Despite this, the EU is now doubling down on its top-down regulatory approach by advancing a massive new regulatory regime for AI, among other new digital regulatory schemes. U.S. tech companies are typically the target of most of these rules since so few major European digital innovators exist.

With the administration’s recent actions, one can’t help but worry that it is looking to follow in the EU’s footsteps on AI policy by implementing more comprehensive controls on computation and meddling in digital tech markets. But there is still time to pursue a more enlightened path. To balance innovation and safety, AI governance must focus on flexible, collaborative, iterative, bottom-up governance solutions through risk-based policies focused on system outcomes rather than system inputs or design.

To achieve a truly safe, secure and trustworthy technological base, the United States must first craft an innovation policy culture that is hospitable to algorithmic entrepreneurialism and investment. We should not forget how the nation embraced and encouraged the internet and digital technology a quarter-century ago with sensible policies that encouraged a massive inflow of talent and capital. This fueled an explosion of world-class tech startups and created a strong technology base that remains the envy of the rest of the world.

While the new EO can help promote positive AI outcomes along certain dimensions, it also opens the door to administrative overreach and bureaucratic micromanagement of a fast-moving set of still-developing technologies with enormous potential to improve human welfare and national security. Prudence and humility should guide AI policy at this stage so as not to derail that potential.