Policymakers in the United States are increasingly focusing on technical standards for artificial intelligence (AI) and other critical and emerging technologies (CETs), especially because they’re accustomed to their firms and their experts being global standards-setters and are concerned about China’s growing participation. They’re concerned about what policy preferences (such as those involving privacy and surveillance) China may embed in technical standards for new technologies at standards bodies, especially those at the International Telecommunications Union (ITU). On February 29, 2024, Senators Warner and Blackburn introduced the “Promoting United States Leadership in Standards Act of 2024.” The bill is well-intentioned, but it needs key changes if it’s to be effective in helping the United States retain its leading role in setting technical standards for CETs. This post analyzes the bill and builds off related analysis of the Biden administration’s strategy for CET standards and how U.S. national security concerns over China shouldn’t imperil its leadership in setting technical standards.

Supporting the U.S. Government and Industry’s Respective Roles in Standards Development

The bill tasks the National Institute of Standards and Technology (NIST) to evaluate and identify opportunities for U.S. government participation in the development of AI and other CET standards for a report to Congress. This is fine, to a degree, but it should be caveated that the primary purpose is to improve U.S. government awareness of what’s happening in key standards bodies. U.S. government officials participate in some standards organizations, but the U.S. government’s direct role in setting standards is largely limited to determining the use of standards for specific government functions that involve national security concerns. The U.S. private sector’s level of engagement and technical expertise will always be orders of magnitude greater than that of the U.S. government, as they’re the ones that create the underlying technologies. The U.S. government does not (with few exceptions) have the technical expertise to develop standards for CETs.

U.S. government engagement in developing CET technical standards should be focused on tracking and following developments rather than having a direct seat at the table to determine the standards themselves. The Biden administration and Congress should focus on setting up public-private mechanisms for meaningful engagement and cooperation (and support for) the private sector. For example, there have been cases where U.S. experts were running for election at 3GPP (an important body developing standards for mobile telecommunications) and the U.S. government didn’t know or didn’t vote in support. The stark contrast between the U.S. government’s and the U.S. private sector’s level of engagement and participation is made clear in instances where U.S. government officials are not even eligible to vote because they have not participated in past meetings (which is often an eligibility requirement to vote).

The one standards body where the U.S. government needs to increase direct engagement is the ITU,as it’s a government membership-based organization, and due to this, is more susceptible to China pushing a coordinated agenda on standards to suit its interests. The vast majority of standards bodies are industry led, and due to their open, transparent, and consensus-based governance, are better protected against any one participant (such as a Chinese firm) trying to unduly influence standards discussions.

The bill should also task NIST and the Department of Commerce to review and propose updates to the United States’ small global network of standards attachés. These attachés, posted at U.S. embassies around the world, are one of the main ways the U.S. government can engage with counterparts on CET standards. The network is well overdue for an overhaul so that it has staff in the right countries (e.g., there should not be a standards attaché in Saudi Arabia) and so that the attaché network’s institutional arrangement relieves staffers of multiple lines of reporting at embassies and with various U.S. government agencies.

Congress and the Biden administration need to know where U.S. federal government agencies are engaged in CET standards and it’s a great idea for the bill to task NIST and the Office of Management and Budget to report U.S. government participation in standards activities. This is a good, but insufficient, step. NIST stopped (with permission from Congress) aggregating and reporting individual agency participation in 2015 (see figure 1). However, NIST never reported which specific standards organizations federal agency staff participated in. The bill should require federal agencies to specify which AI and CET standards bodies U.S. government officials participate in. There are thousands of governmental, quasi-governmental, and private sector/non-governmental standards bodies working on all manner of technology issues. NIST and other government agencies only have limited capacity to participate, so it’s important that Congress and the administration know which ones they should prioritize and participate in.

Figure 1: U.S. Federal Agency Participation in Private-Sector Standards Bodies (2015)

The bill’s tasking of NIST should go further. NIST should work with the State Department, the Department of Commerce, and other agencies and U.S. private sector stakeholders to identify where U.S. stakeholders face barriers to participation. To better address China’s sometimes problematic standards behavior at the ITU, agencies should advise Congress on how they can support industry participation at the ITU, as China, Russia, and others have waged a long war to sideline and exclude the private sector representatives as they frustrate their efforts to push through their preferred standards. However, China,Russia, and the ITU are not the sole threats to the open standards-setting system and U.S. leadership in technical standards. The EU continues to enact measures to marginalize and exclude U.S. experts from European standards bodies and standards-setting activities, including on AI, as part of an effort to use technical standards for protectionism.

Greater Pre-Standardization Cooperation on CETs is a Smart Idea

NIST’s report to Congress also must identify where pre-standardization cooperation would support U.S. leadership in setting CET technical standards, especially via meetings in the United States. This is a great idea. The U.S. government should work with U.S. industry stakeholders to identify which CETs would benefit from public-private pre-standardization discussions on CETs. The challenge is identifying CETs that are at the appropriate (early) stage of maturity where pre-standardization cooperation would be beneficial. Pre-standardization cooperation is a creative, forward-looking way for the United States and its partners to work together on the early alignment of key terminology and concepts, measurement, technical details, use cases, and policy discussions, which can feed into formal standards development.

One doesn’t want multiple, conflicting foundational elements on new technologies around the world. For example, no one wants 70 different definitions, concepts, and measurement specifications for AI. In contrast, it can be hard to achieve this for mature technologies, as the technologies are built out and often tied to political objectives, built up infrastructure, sunk investment costs, and laws and regulations. Pre-standardization cooperation accelerates formal standards development processes, as respective stakeholders are already aligned on foundational issues related to new technologies. This would build on prior successful U.S.-international pre-standardization cooperation on advanced materials, coarse particle content, cloud computing, and nanotechnology.

Supporting More Standards Meetings in the United States: Funding Is Only Part of the Solution

The bill’s centerpiece is a NIST pilot program to support U.S.-hosted meetings of CET standard-setting bodies. NIST has been authorized to spend up to $10 million on the pilot project, but no new funds have been provided (see below). NIST would provide grants to cover the costs of preparing and holding meetings, but not exceeding half of the total cost. The bill details criteria for NIST to evaluate grant applications, such as whether the organization is involved in AI and CET standards setting, has a demonstratable history of hosting successful meetings, and whether it has a stable or growing participant base. While this is a worthy idea, the problem is that Congress has a poor track record of appropriating extra funds for NIST, even as it and the administration expands NIST’s agenda on CETs (see below).

However, without addressing the U.S.’s problematic visa adjudication and approval process, hosting more meetings in the United States is two steps forward, one step back as U.S. participants can more easily participate, but not officials from China, India, and elsewhere. The bill does nothing to study or address the issue of how to make it easier for foreign experts working in standards bodies to get U.S. visas. Participants from other countries often experience delays and an unpredictable visa application process that leads standards bodies to hold meetings in countries with more accessible and predictable visa processes.

Blocking these officials from attending standards meetings in the United States is counterproductive. U.S. visa issues spur these experts and international standards bodies to hold their meetings elsewhere. It also makes it easier for them to argue to their home governments (and standards bodies) that they should develop their own separate standards because they are barred from providing input and discussing the international standard.

This is what happened when U.S. sanctions did not include a clear and broad exemption for targeted entities like Huawei to continue participating in standards-related activities (the Biden administration eventually fixed the issue). Standards bodies that included participants from Huawei and other targeted firms left or avoided the United States for fear of breaching U.S. sanctions laws. Chinese experts went home, and with the help of the Chinese government, setup duplicative and objectively worse local standards that disadvantage U.S. firms, technology, and products.

Another tangible way to support U.S. participation at CET standards meetings would be for the bill to modify the research and experimentation tax credit to allow international standards setting costs to qualify as expenditures for purpose of the credit. It is costly for companies, especially small- and medium-sized companies, to participate in international standards discussions. One estimate is that it can cost a company up to $300,000 per year for an engineer to work on international standards.

NIST: Already Overworked and Underpaid

The bill’s core weakness is the lack of appropriated funding for NIST for the pilot project and this other new work. So, either NIST must come up with the money from its existing budget or it can simply not do the project. This is becoming a common challenge for NIST. In a recent op-ed, Congresswoman Zoe Lofgren (D-Calif, ranking member of the House Committee on Science, Space, and Technology) criticized these budget cuts as part of the recently approved mini-appropriations bill.

NIST has been tasked with a growing agenda (including the AI Safety Institute) for which new funds have not been appropriated. What was not covered in recent reporting about NIST staff reportedly threatening to quit over appointments to the AI Safety Institute was that besides the lack of appropriated funds, NIST also has to shell out close to $300M in the form of earmarks. NIST’s budget is being hit from all sides. For example, operating under a continuing resolution means NIST’s quarterly spending consistently falls short of requirements. On top of this, with cost-of-living allowance (COLA) increases (which went into effect on January 1, 2024), NIST’s budget takes another hit as it must cover this from elsewhere in the budget.

Given it’s an election year and NIST is getting funds for CHIPS and Science Act programs, there is very little appetite (unfortunately) in Congress to give NIST more funding for AI and other CETs, even with the formation of the NIST AI Safety Institute. This means that until NIST gets more funding, these types of standards-specific bills (of which there’s been dozens in recent years) actually make the situation worse for NIST, as it forces them to do even more with less.

Conclusion: If Funded and Revised, the Law Will Support U.S. Standards Power

True standardization influence comes from contributors submitting early, well-developed drafts that form the basis for discussions that ultimately lead to final standards. The expert (or experts) that drafts first, and drafts well, has influence. The United States has developed a pool of mainly private sector experts that are good at this. The U.S. government should clearly do more to help them and to ensure government officials are better able to track developments at key CET standards bodies. The bill includes some useful ideas to support U.S. leadership in CET standards, but it could go further. Hopefully these shortfalls are addressed and this bill (unlike the many other standards-focused bills) becomes law and thus supports the U.S. government’s increasing focus on AI and CET standards.