Elizabeth Hyman

Last week’s Supreme Court's decision to overturn the #Chevron doctrine has sent ripples through the regulatory and #policy communities, especially in the world of healthcare, particularly concerning #AI. This landmark ruling limits federal agencies' authority to interpret ambiguous laws, potentially complicating efforts to oversee rapidly evolving technologies like AI, something my colleague Joseph Hoefer and I have been tracking closely. Justice Elena Kagan, in her dissenting opinion, raised alarm bells about the decision's impact on regulating complex fields such as AI, warning that it could hinder Congress's ability to delegate regulatory power to federal agencies effectively. The ruling presents a significant challenge for lawmakers and regulatory bodies like the FDA, which have been working on programs to regulate health AI. These efforts, including draft guidance for AI/ML-enabled device software, may now face new hurdles under the changed legal landscape. Lawmakers are already at a crossroads, needing to decide whether to regulate health AI through existing agencies or create a new body dedicated to AI oversight. The Chevron ruling emphasizes the need for Congress to be more specific and prescriptive in future legislation, potentially delaying much-needed oversight of advancing technologies. The majority opinion, authored by Chief Justice John Roberts, argued that deferring to regulators under Chevron is inconsistent with the Administrative Procedure Act. However, dissenting justices contend that this ruling puts courts at the apex of the administrative process across all subjects, potentially overreaching their intended role. As AI continues to integrate into various aspects of healthcare, the implications of this ruling will likely be felt for years to come. The decision could significantly slow down efforts to establish necessary guardrails for AI in #healthcare, leaving the industry in a state of uncertainty as it grapples with how to ensure responsible development and deployment of these powerful technologies.

32

For the third installment of our AI and trade collaboration at Monument Advocacy, Poorvie Bishnoi, Caroline Reppert, and I delve into how the Administration and Congress are positioning to leverage export controls on AI to safeguard American interests. The Biden Administration and bipartisan lawmakers look to continue to impose stricter export controls on cutting-edge AI technology that could make its way to US adversaries like the People’s Republic of China (PRC), who may use these tools to conduct cyber-attacks, spread disinformation, or develop biological weapons. A new bill – the ENFORCE Act – recently introduced by House Foreign Affairs Committee Chairman Michael McCaul (R-TX) and Rep. Susan Wild (D-PA) along with Select China Committee Chairman John Moolenaar (R-MI) and Ranking Member Raja Krishnamoorthi (D-IL) - would give the Commerce Department expanded authority to restrict exports of the most advanced AI models. Crucially, it removes legal hurdles around regulating open-source models that can be freely downloaded from the internet – something China has heavily relied on to advance technologically. The proposed legislation would restrict the export of certain AI models – regardless of whether they are open source or proprietary – and bar certain American individuals and companies from working with foreign entities on developing high-risk AI capabilities. In parallel, the Commerce Department is considering new rules to restrict exports of proprietary or closed-source  AI models not currently subject to oversight. This would complement existing curbs on exporting advanced AI chips and tools to rival nations like China. Regulators may use thresholds based on the computing power required to train AI models as criteria for determining which ones require export licenses. However, intended use cases, like protein design for potential bioweapons development, could also be evaluated. The moves come as the US #intelligence community is sounding alarms about hostile actors leveraging advanced AI #technology to deploy #cyberattacks. There are also concerns that adversaries could misuse language models to spread #disinformation, destabilize governments, or develop biological weapons. With AI rapidly progressing, #export controls are vital to safeguarding American interests from adversaries seeking to undermine them. However, any legislation and potential new rules are likely to renew debates around striking the right balance between protecting national security and maintaining international research collaboration. Managing that trade-off will be a key challenge as artificial general intelligence nears. As an #AI arms race intensifies, enforcing #guardrails without stifling innovation and international cooperation will require a calibrated and collaborative approach from policymakers. 

16

1 Comment

For anyone interested in organizational culture - this important session will go beyond the hype and hyperbole and address not just how AI will affect what we do and how we do it, but how we work together.

8

This article serves as your regular reminder that every agency is looking to become an AI enforcement agency. Samuel Levine, the FTC's Director of the Bureau of Consumer Protection, argues that "Congress giving us back that tool [restitution] to get money back to people who get scammed, and making sure we have the resources to take on what we fear will be a real scourge of AI-related fraud, is essential to us doing our job." This has been a pretty consistent argument from the FTC, ever since a unanimous Supreme Court ruled in 2021 that Section 13(b) of the FTC Act did not authorize use of that provision to seek restitution. But that "giving us back" is doing a lot of work. As Justice Breyer wrote in AMG Capital Management, "We must conclude... that §13(b) ... does not grant the Commission authority to obtain equitable monetary relief." In other words, the FTC has never had restitution authority under §13(b). There is nothing to "give back." That issue aside, any company making claims about the capabilities of their AI systems are on notice that the FTC is watching. This is not a "Big Tech" issue - any company, small or large, can be held liable for violations of the FTC's consumer protection authority. And companies entering into settlements with the FTC are often subject to a 20-year consent decree. #ai #artificialintelligence #federaltradecommission #ftc

7

Flagging Federal Registry notice seems akin to tuning into a Congressional Roll Call on C-SPAN, but this is the State Department’s attempt to lean on the #privatesector, #academia, and #thinktanks by establishing Federally Funded Research and Development Centers (FFRDC). Written comments are due on August 15, 2024, at 5 PM ET, and I hope we can benefit from broad experience and insights to set these up to modernize our systems, operations, and the field of diplomacy. The goal is to create mechanisms for the State Department to benefit from deep technical research that isn't normally associated with the field of #diplomacy and #foreignaffairs. #doingbusinesswithdepartmentofstate #modernization

36

1 Comment

This resonates with the findings from our joint Mathematica & @Results4America report: 1. Cities with strong results-driven contracting practices developed Recovery Plans that better adhere to federal guidance for key data and evidence practices. 2. Cities that effectively engage residents and local organizations strengthen the impact of federal recovery funds by continuing to use data to identify and embed equity. 3. Cities that have been recognized as effectively using data by achieving What Works Cities Certification were better positioned to use data and evidence practices to advance economic recovery, economic mobility and racial equity through State and Local Fiscal Recovery Funds funds. Read more on "Unlocking the Power of City Data: Transforming American Rescue Plan Investments to Improve Residents’ Lives" here: https://lnkd.in/g5MHcPVh.

8

Join Rob Berlin, Lauren Oliva, PharmD, Julie Tai, Jeffrey Siegel and me as we discuss " Challenges in exceeding the quality of existing endpoints and approaches using digital tools" on Tuesday, June 18 (session 223). Lets discuss whether the advantages of using digital tools to develop novel measures can be adequately validated using existing "gold standards" or whether it will dilute their potential. I'll speak about our publication "Incorporating digitally derived endpoints within clinical development programs by leveraging prior work" in #naturedigitalmedicine which builds on #DiME's V3 framework and much more. #patientfocusseddrugdevelopment #digitalhealth #DIAglobal2024

76

4 Comments

The NEPA reform rule mandated in the Bipartisan Infrastructure Law came out today. Last week, the EPA released its much anticipated regulations limiting emissions from power plants. Also last week, the federal government banned employer noncompete agreements. AND there was a rule requiring airlines to provide cash reimbursement for some delayed flights and mandating transparent flight pricing. And the rush will continue for most of May. Why? The Biden Administration has until about May 22 to issue rules that will be difficult to overturn. Any rule issued after that date could be impacted by the Congressional Review Act, which enables a president and a simple majority in Congress to overturn any major rules issued within the last 60 days of a Congressional session. (Of course, if Biden wins in November, the CRA is less of a risk--but an administration always wants to play it safe with the CRA.) So buckle up, grab your favorite caffeinated beverage, and buy a fresh pair of reading glasses. It's time to read the Federal Register. More of an explainer from Reuters in the comments. #Regulations #Politics #USElections #Sustainability

8

1 Comment

Today, the Department of Homeland Security announced the establishment of the Artificial Intelligence Safety and Security Board to advise the secretary on critical infrastructure risks. While a worthwhile initiative, the design of our current critical infrastrucure policy structure is a fundamental disservice to such efforts. Today, critical infrastrucure encompasses 16 industrial sectors, an unwieldy 60% or more of the U.S. economy, and a grab bag of non-critical assets including casinos, ride share vehicles and the cosmetics industry. For a small committee of 22, this unwieldy scope will strain priorities, waste time, and distract focus from the core systems that really matter. Effectively securing our critical infrastrucure demand we set the priorities, shrink policy scope, and finally determine what is indeed the most critical and the most deserving of DHS’s limited resources. https://lnkd.in/eXdRveWB

5

1 Comment

My latest in the The Federalist Society blog: Some Thoughts on NTIA's New "Artificial Intelligence Accountability Policy Report" #NTIA #AI #ArtificialIntelligence https://lnkd.in/eu_8rFj6

2

The US Patent and Trademark Office (USPTO) has recently released a new National Strategy for Inclusive Innovation: https://lnkd.in/gUWPBeNh There are deep seated inequities in the innovation ecosystem and the ability for people from diverse backgrounds to both become innovators and to benefit from innovation. Recent analysis by Statistics Canada | Statistique Canada and Canadian Intellectual Property Office / Office de la propriété intellectuelle du Canada, for example, highlights that from 2015 to 2019 only 11.9% of inventors are women. https://lnkd.in/g5QNe2W2 It is interesting to see how USPTO are proposing to tackle some of the barriers that exist from an IP angle.

3

Amid the challenges posed by the Israel/Gaza demonstrations on American campuses, along with a rising tide of political attacks on the autonomy of American universities, shared governance is also eroding on many campuses. The internal conflict that is splitting university communities, and fueling outside criticism, is also lending momentum to university presidents, and governing boards, ignoring and downgrading the voice and role of faculty senates – to some degree joining forces with lawmakers in some states who are attacking tenure, forcing curricular changes, demanding the end to the nomenclature of DEI, and view faculty as the problem. See Ryan Quinn’s recent InsideHE article which discusses the desire of some faculty at Harvard to establish a greater role in university management in the wake of the Claudine Gay resignation, as well as a scan of legislation demoting the role of faculty in university affairs in states like Florida, Arizona and West Virginia. https://lnkd.in/g_NQNudj See also my article on the critical role of shared governance in developing the University of California’s multi-campus system. https://lnkd.in/gt8YnzRS #universities #faculty #Israelgaza #sharedgovernance

6

The Bipartisan Senate #AI Working Group unveiled their AI Roadmap. Their focus areas are spot on: fostering #innovation, enhancing #workforce development, and implementing a risk-based approach to #AIgovernance and #responsibleAI. In my latest blog, I explore the key elements of this roadmap, including: 🤝 Promoting public-private partnerships to mitigate AI risks. 🔩 Securing funding for AI initiatives at the National Institute of Standards and Technology (NIST) 🎓 Creating incentives for employers to integrate #technology and #reskilling, and for workers to pursue #retraining.

50

3 Comments

🚨 The Department of Justice Antitrust Division opened a portal for individuals to submit a heath care competition complaint. While I'm not super hopeful that this will result in any major action against vertical integration in the health care marketplace, it's a step, if just a baby one. #healthcare #healthpolicy #congress #verticalintegration #competition #health ✉ Get your comments in!

1

Still, the response to AI from both policymakers and Congress requires far more urgency. We must discard the often-unstated presumption that an AI-driven solution is necessarily preferable or inevitable. Instead we must ask ourselves, as a threshold question, whether an AI tool or model is sufficiently accurate, accountable, transparent, fair, and safe to be used in this way. And we must ensure that deployers of AI tools—in both government and the private sector—understand, train for, and communicate publicly about limitations and biases. — From my remarks to the NAIAC yesterday (full text at link below)

10

If you are a company or organization currently paying or looking for government relations, public affairs, or lobbying help in the Washington, DC area, don't pay the additional premium for the firm with the unnecessary office image in DC to support. Doing so is like Jim Croce continuing to put dimes into a pay phone to keep the connection going with the operator while a growing digital era takes hold. Instead look at smaller firms, like Avison Strategies, LLC, with similar experience, a much smaller fee structure and a business model focused on results, the latest AI and ML technology, and personal attention to reach people where they are and when it is needed. Retainers average $4,000-$5,000/month for firms like mine compared to $10,000-$15,000/month for firms like those. Office costs and the impact through association fallacy will make you overpay and results will frequently under deliver. Support a small business instead and see the difference.

1

I'm keeping an open mind about the potential for AI to impact CBRN threats and generally encourage efforts to measure these issues in order to ground the conversation and minimize often-fantastical dooms-day scenarios. That said, evidence that AI is meaningfully changing current CBRN threat levels is lacking and probably overstated. Beyond this report, DHS must consider how to best prioritize its limited AI bandwidth. With a fixed budget, any excess money spent on mitigating or analyzing these hypotheticals nessesarily comes at the expense of other, evidence-backed priorities. Rather than over-invest in AI-CBRN issues, DHS should spend on the following: 1. Cybersecurity and other evidence-based challenges: while early days, there is indeed evidence that AI has changed the cyber game. In the past year spear phishing increase 1,200%, machine translation improved the quality of social engineering attacks, and state backed actors have used LLMs to generate operational malware. While not quite a crisis, this is a meaningful change that could portend a real challenge if capabilities continue to improve. Rather than chase CBRN 'what ifs,' resources should focus on this very real, evidence-backed challenge. 2. Preparing for the unknown: Its still early days, the tech is changing rapidly, and its uncertain what challenges will be the core AI issues of the coming years. Rather than overinvest resources on CBRN threats, DHS should invest in the agility and capacity needed to quickly identify and react to a range of unexpected, emergent challenges. The first step is state capcaity, DHS should focus on upskilling and hiring to build a work force that not only understands AI, but AI's intersection in a broad diversity of domains. Second is regulatory flexibility, DHS should actively analyze its rules to ensure it can act with minimal bureaucratic delay if we see a black swan AI development. Third is vigilance, DHS should foster the infrastructure, such as Information Sharing and Analysis Centers, needed to quickly identify unexpected changes in the threat environment. It is indeed worthwhile for the Countering Weapons of Mass Destruction office to analyze how AI could impact their work. Without evidence, however, this shouldn't become a preoccupation. As the DHS proceeds it should focus limited bandwidth on proven challenges and for all else, build the threat-agnostic capacity needed to act with agility.

7

(Our tutorials are out @ https://lnkd.in/gTVe5W9b - check us out!) "Yong Kiat, why haven't you been updating me on SENSE more regularly?" I wish I could update stakeholders and Senior Management on our products the way I did when I was a MOE or MOH policy officer - daily, or even weekly. But the truth is that in modern software engineering, realities and challenges change so frequently, that it's impossible for anyone to have a firm grasp on the product evolution other than the product team. And that's why the 21st century public servant has taken to LinkedIn to place strategic updates. Don't worry, we've been hard at work! As part of product launch, we've prepared twelve tutorial videos on how to operate SENSE, and its various LLM features! We translated prompt engineering techniques like context creation, context clarification and few-shot prompting into practical queries to ask on SENSE. Why tutorials? Well, a great product is nothing if we do not make it clear as day how our product should be used. Here at GovTech, we're not just interested in creating products for the public service. A digital savvy workforce is as much the people as the digital products they use - since LLMs are such a novel tool, we thought to place more emphasis on education and good product design, rather than just ride the AI product bandwagon. This will pave the way for, well, a sensible use of SENSE. Thanks for partnering us on our journey, and being with us during this exciting launch month for SENSE.

49

5 Comments