WASHINGTON – U.S. Sens. Mark R. Warner (D-VA), Chris Coons (D-DE), Jeanne Shaheen (D-NH), Jack Reed (D-RI), and Mark Kelly (D-AZ), as well as Congressmen Jim Himes (D-CT) and Raja Krishnamoorthi (D-IL) issued the following statement in response to President Trump’s artificial intelligence deals that were announced with the United Arab Emirates and Saudi Arabia this week:

“Democrats and Republicans have long agreed that American companies must remain the undisputed leader in AI, a rapidly developing technology critical to the future of everything from our national security to manufacturing, finance to health care. We have worked hard to ensure the most powerful AI systems are built here, and we have fought to restrict the most sophisticated chips from reaching China – or those who would grant remote access to China – given Beijing’s use of AI to strengthen its military, crack down on domestic dissent, and compete with the U.S.

“President Trump announced deals to export very large volumes of advanced AI chips to the UAE and Saudi Arabia without credible security assurances to prevent U.S. adversaries from accessing those chips. These deals pose a significant threat to U.S. national security and fundamentally undermine bipartisan efforts to ensure the United States remains the global leader in AI. Rather than putting America first, this deal puts the Gulf first.

“The volume of AI chips Trump is offering for export would deprive American AI developers of highly sought-after chips needed here and slow the U.S. AI buildout. Under this deal, data centers and AI systems that would otherwise be built in America will be built in the Middle East – at the exact time that President Trump says he wants to bring jobs and key industries back home. This deal would incentivize U.S. firms to build the factories of the future overseas, creating significant vulnerabilities in our AI supply chain. If our leading AI firms offshore their frontier computing infrastructure to the Middle East, we could become as reliant on the Middle East for AI as we are on Taiwan for advanced semiconductors – and as we used to be on the Middle East for oil. We should not foster new dependencies on foreign countries for this premier technology.

“Additionally, these deals will provide our highest end chips to G42, a company with a well-documented history of cooperation with the People’s Republic of China. We applaud the administration's efforts to limit exports of advanced AI chips to China, including recent actions to further restrict exports of Nvidia chips. However, these efforts will be for nothing if G42 or other companies with ties to China are given large quantities of our most advanced chips. 

“Proponents of the deal argue that China will fill the gap if we do not sell substantial quantities of advanced chips to these countries. This is false. China cannot and will not because China makes fewer chips as a nation than these deals offer, and each is inferior to their U.S.-designed equivalent. This is thanks to the bipartisan efforts under both the Trump and Biden administrations to cut off China’s access to advanced chip manufacturing equipment. These efforts have worked, and we should double down on this success rather than squander the leverage we have won.

“If this deal succeeds, the offshoring of frontier American AI will be recorded as an historic American blunder. People around the world deserve to enjoy the benefits we will reap from AI. However, AI chips must only be exported to trusted companies, in reasonable numbers, and in concert with credible security standards and assurances. We welcome the opportunity to work with the administration to meet these objectives and urge our colleagues in Congress to do the same.”

Senator Warner is Vice Chair of the Senate Intelligence Committee. Senator Coons is Ranking Member of the Senate Appropriations Subcommittee on Defense. Senator Shaheen is Ranking Member of the Senate Foreign Relations Committee. Senator Reed is Ranking Member of the Senate Armed Services Committee. Senator Kelly is a member of the Senate Intelligence Committee. Congressman Himes is Ranking Member of the House Intelligence Committee. Congressman Krishnamoorthi is Ranking Member of the House Select Committee on the Chinese Communist Party.

###

U.S. Sen. Mark R. Warner (D-VA), Vice Chairman of the Senate Select Committee on Intelligence, urged the leaders of federal departments and agencies to promote data collection and transparency around their adoption of artificial intelligence (AI). In a series of letters to 23 department and agency heads, he emphasized the critical importance of collecting data on AI’s impacts promoting productivity and improving government outcomes, and he posed a series of questions about how agencies are making decisions that weigh the benefits and risks of the technology alongside the experiences of federal employees.

AI is having a profound impact on the workforce across both public and private sectors, often allowing workers to complete tasks more efficiently and cost-effectively. In Fiscal Year 2022, 20 of 23 federal agencies reported almost 1,200 current and planned AI use cases. However, agencies haven’t been forthcoming with data on how AI has changed outcomes for federal agencies, or how data is being used to inform future decision-making.

The letter highlights several of these potential and actual use cases, saying, “Per the AI case use inventory, the utilization of artificial intelligence across federal departments and agencies has allowed the federal workforce and contractors to work efficiently and creatively – improving government operations and delivering better results for the American people. These examples include the Social Security Administration using AI to expedite determinations for disability benefits, the Department of Veterans Affairs utilizing AI to capture trends and facilitate processing of veteran feedback, and the Department of Justice applying AI to accurately identify and process threat tips.”

In the letter, Sen. Warner stressed the urgent need for departments and agencies to release data on their AI adoption, so they can implement best practices and work to deliver better results for the American people. It also asks how feedback from federal employees and contractors is being considered to make decisions that best support both workers and outcomes.

“While government-sourced, publicly-available information provides sector or task-specific summaries of how the aforementioned federal departments and agencies are adopting artificial intelligence, I am concerned about the limitations of this information with respect to the broader adoption at scale of AI in the federal government, including the need for measurable data and conclusive assessments on how individual AI use cases are enhancing the missions of federal departments and agencies,” Sen. Warner continued. Establishing data collection standards that track the progress of AI’s adoption in the federal government will help better understand the state of integration, assess its effectiveness, implications, and appropriate usages, and guide the direction of future adoption plans.”

Sens. Warner sent the letter to the following federal departments and agencies: The Departments of: Agriculture, Commerce, Defense, Education, Energy, Health and Human Services, Homeland Security, Housing and Urban Development, Justice, Labor, State, Interior, Transportation, Treasury, and Veterans Affairs, and the Environmental Protection Agency, the General Services Administration, NASA, the Nuclear Regulatory Commission, the National Science Foundation, the Small Business Administration, the Social Security Administration, and the U.S. Agency for International Development.   

Sen. Warner, a former tech entrepreneur and Vice Chairman of the Senate Select Committee on Intelligence, has been a longtime leader on AI-related issues, particularly within the federal government. He led bipartisan legislation to help the federal government mitigate risks associated with AI while still being able to benefit from this emerging technology. In May 2024, he introduced bipartisan legislation to improve the tracking and processing of security and safety incidents and risks associated with AI, including through improving information sharing between the federal government and private companies. He also has repeatedly pushed on companies to keep their promises to promote security and safety throughout the rollout of novel AI technologies.

A copy of the letter is available here and below:

I write to you today regarding your agency’s utilization of artificial intelligence (AI) systems and enabled technologies and request information on your department or agency’s use of those systems and technologies. I request sufficient information to understand the purposes to which your department or agency uses those systems, the analyses of the possible and actual uses of AI applications by your department or agency, and the metrics by which your department or agency evaluates the use of those systems, including by federal workers and contractors.

In a world where we are still working to understand the full capabilities and impact of advancements in artificial intelligence, it is critical that the federal government lead in data collection and evidence-based decision-making in the adoption of these technologies. In that same vein, the adoption of AI tools by the federal government should be based on measurable outcomes, such as productivity gains.

The use of artificial intelligence across various occupations and industries is transforming the labor market and impacting the global economy broadly. More specifically, the application of AI in the workforce has yielded promising results, including the potential for increased worker productivity. In many instances, artificial intelligence has allowed for tasks to be completed faster and more efficiently, allowing workers to focus on high-value responsibilities and expanding their range of work.

The private sector, particularly innovative artificial intelligence companies and the businesses that use their products, are leading the charge in measuring and providing real-time dynamic data on the impact of artificial intelligence technologies on their workforce and worker productivity. This data includes, but is not limited to, specific measurements on how AI has led to time-saved on specific tasks, production volume, improving error rates, and customer satisfaction. These metrics and subsequent analyses are useful in evaluating the impact and value of artificial intelligence.

As of the 118^th Congress, the federal government employs over 2 million individuals, with the Commonwealth of Virginia holding the third-largest constituency of federal civilian employees. For FY2024, the federal government executed over 104 million contracts, similarly employing, directly and indirectly, individuals to carry out the missions of federal departments and agencies. These public servants perform essential work for our country, and as detailed below, some of their work is complemented by and supplemented through the integration of artificial intelligence systems and technologies.

The Government Accountability Office (GAO) demonstrated that in FY2022, twenty of 23 agencies reported about 1,200 current and planned artificial intelligence use cases. Per the AI case use inventory, the utilization of artificial intelligence across federal departments and agencies has allowed the federal workforce and contractors to work efficiently and creatively – improving government operations and delivering better results for the American people. These examples include the Social Security Administration using AI to expedite determinations for disability benefits, the Department of Veterans Affairs utilizing AI to capture trends and facilitate processing of veteran feedback, and the Department of Justice applying AI to accurately identify and process threat tips. The use case inventory applies the definition of artificial intelligence as provided in Section 238(g) of the John S. McCain National Defense Authorization Act for Fiscal Year 2019.

While government-sourced, publicly-available information provides sector or task-specific summaries of how the aforementioned federal departments and agencies are adopting artificial intelligence, I am concerned about the limitations of this information with respect to the broader adoption at scale of AI in the federal government, including the need for measurable data and conclusive assessments on how individual AI use cases are enhancing the missions of federal departments and agencies. Establishing data collection standards that track the progress of AI’s adoption in the federal government will help better understand the state of integration, assess its effectiveness, implications, and appropriate usages, and guide the direction of future adoption plans.

As such, I respectfully request that you respond to this letter with detailed answers to the following questions by January 17, 2025: 

• Does your department or agency utilize AI?
  • If yes, please provide a high-level summary of the utilization of AI, including uses by administrative or operational units of your department or agency.
  • If no, please detail how your department or agency reached the decision to not utilize AI.
• How does your department or agency identify uses cases, needs, or other instances in which it deems the use of AI to be appropriate? Please describe in detail the decision-making process that your department or agency has, does, or plans to undertake when determining if the use of AI is appropriate.
• Regarding future or planned uses of AI, how does your department or agency incorporate data collection and identify measurable outcomes when determining if the use of AI is appropriate? What metrics does your department or agency use when determining the appropriateness of AI?
• Regarding current uses of AI, how does your department or agency incorporate data collection and identify measurable outcomes when determining if the use of AI is productive or effective? What metrics does your department or agency utilize when determining the productivity or effectiveness of current AI applications? How do these metrics and data collection guide decision-making on future applications of AI?
• Does your department or agency measure worker productivity or productivity gains as a result of the application of AI?
  • If yes, please detail how your agency measures worker productivity. How does this guide your department or agency’s decision-making on future applications of AI?
  • If no, please detail why your agency does not measure this.
• Please describe in detail the process that your department or agency uses to solicit input or feedback from the federal workers or the contractors who will be directly utilizing the planned AI technology.
• When determining if the use of AI by your department or agency is appropriate, please describe in detail how your department or agency considers the need for additional training for the federal workers and contractors who will be directly applying the AI technology as part of their job duties and responsibilities.
• If your department or agency is utilizing AI, please describe in detail how those uses inform your department or agency’s considerations on adjusting mission approach or allocating tasks among the department or agency’s workforce, including, but not limited to, adjusting job responsibilities, daily tasks, or team compositions?

I appreciate your thoughtful consideration of this matter and look forward to your response.

WASHINGTON – Today, Senate Select Committee on Intelligence Chairman Mark R. Warner (D-VA) issued the following statement in response to President Biden’s National Security Memorandum (NSM) on Artificial Intelligence:

“As we have seen just over the last two years, AI technology is rapidly evolving in a way that will have massive consequences for our economy, national security, and even democracy. I am heartened to see the administration recognize this very fact and take a leadership role to advance AI capabilities while simultaneously promoting responsible research, strong governance that ensures trust and safety, and the protection of human and civil rights.

“I am also gratified to see that the NSM appears to implement many of the legislative proposals I have advanced, including requirements to promote AI security research and address AI cyber vulnerabilities. However, as the chair of the Senate Intelligence Committee I am also acutely aware of the many threats to our AI efforts. I encourage the administration to work in the coming months with Congress to advance a clearer strategy to engage the private sector on national security risks directed at AI systems across the AI supply chain.” 

###

WASHINGTON – U.S. Sen. Mark R. Warner (D-VA), Chairman of the Senate Select Committee on Intelligence and co-chair of the Senate Cybersecurity Caucus, wrote to the U.S. Copyright Office in support of expanding the existing good-faith research exemption within the Digital Millennium Copyright Act (DMCA).

Every three years, the DMCA goes through a rulemaking process to authorize exemptions that allow individuals and researchers to circumvent technical protection measures on copyrighted material without risking liability. This year, Artificial Intelligence (AI) researchers have petitioned for a new exemption relating to “Security Research Pertaining to Generative AI Bias.” Sen. Warner, who has led the charge in the Senate to explore the capabilities of AI technology while simultaneously advocating for reasonable guardrails around its usage, argues that expanding the current good-faith research exemption to cover research that falls outside of traditional security concerns, such as bias and other harmful outputs, is the best way to ensure safe and equitable AI while enabling its continued innovation, public trust, and adoption.

Sen. Warner wrote, “Due to the difficulty in understanding the full range of behaviors in AI systems - particularly as models are introduced in contexts that diverge from their intended use - the scope of good-faith research has expanded to the identification of safety flaws caused by misaligned AI systems, as well as research into how AI systems can reflect and reproduce socially and economically harmful biases…it is crucial that we allow researchers to test systems in ways that demonstrate how malfunctions, misuse, and misoperation may lead to an increased risk of physical or psychological harm.”

He continued, “At the same time, as the Department of Justice letter emphasized, a hallmark of the research exemption has been the good faith of security researchers. In the absence of regulation, many AI firms have voluntarily adopted measures to address abuse, security, and deception risks posed by their products. Given the growing use of generative AI systems for fraud, non-consensual intimate image generation, and other harmful and deceptive activity, measures such as watermarks and content credentials represent especially important consumer protection safeguards. While independent research can meaningfully improve the robustness of these kinds of authenticity and provenance measures, it is vital that the Copyright Office ensure that expansion of the exemption does not immunize research that intends to undermine these vital measures; absent very clear indicia of good faith, efforts that undermine provenance technology should not be entitled to the expanded exemption.”

This is the latest step in Sen. Warner’s efforts to reign in big tech and better understand the impacts of rapidly expanding usage of AI. Earlier this month, he introduced the Secure Artificial Intelligence Act of 2024, legislation to improve the tracking and processing of security and safety incidents and risks associated with Artificial Intelligence (AI).

A copy of the letter is available here and below: 

Dear Ms. Perlmutter,

I write today in response to the petition submitted to your office that proposes a new exemption for “Security Research Pertaining to Generative AI Bias” as part of the Copyright Office’s ninth triennial rulemaking proceeding under the Digital Millennium Copyright Act (DMCA). I understand a number of stakeholders have submitted public comments to weigh in on this petition, including a letter from the Department of Justice. Ultimately, I urge the Copyright Office to consider expanding the existing good-faith security research exemption to cover both security and safety flaws or vulnerabilities, where safety includes bias and other harmful outputs.

As the leader of bipartisan legislation to improve the security of AI systems and the Co-Chair of the Senate Cybersecurity Caucus, I recognize the importance of independent security research. The existing DMCA exemption for good-faith security researchers plays a critical role in empowering a robust security research ecosystem that identifies vulnerabilities and risks to systems around the world, facilitating their remediation, and preventing future exploitation by threat actors that could lead to incidents. We must continue to promote this important work and understand that, although AI is software at its core, the non-deterministic nature of AI systems means that security vulnerabilities are no longer the only type of flaw that can be introduced and enable misuse. As the AI Risk Management Framework, developed by the National Institute of Standards and Technology (NIST), emphasizes, AI risks differ from traditional software risks in key ways - including increased opacity and barriers to reproducibility, complex and non-deterministic system dependencies, more nascent testing and evaluation frameworks and controls, and a “higher degree of difficulty in predicting failure modes” for so-called “emergent properties” of AI systems.

Due to the difficulty in understanding the full range of behaviors in AI systems - particularly as models are introduced in contexts that diverge from their intended use - the scope of good-faith research has expanded to the identification of safety flaws caused by misaligned AI systems, as well as research into how AI systems can reflect and reproduce socially and economically harmful biases. This research into bias and other harmful outputs is essential to ensuring public safety and equity while enabling continued innovation, public trust, and adoption of AI. Therefore, it is crucial that we allow researchers to test systems in ways that demonstrate how malfunctions, misuse, and misoperation may lead to an increased risk of physical or psychological harm.

At the same time, as the Department of Justice letter emphasized, a hallmark of the research exemption has been the good faith of security researchers. In the absence of regulation, many AI firms have voluntarily adopted measures to address abuse, security, and deception risks posed by their products. Given the growing use of generative AI systems for fraud, non-consensual intimate image generation, and other harmful and deceptive activity, measures such as watermarks and content credentials represent especially important consumer protection safeguards. While independent research can meaningfully improve the robustness of these kinds of authenticity and provenance measures, it is vital that the Copyright Office ensure that expansion of the exemption does not immunize research that intends to undermine these vital measures; absent very clear indicia of good faith, efforts that undermine provenance technology should not be entitled to the expanded exemption.

The existing exemption has been an important contributor to the multistakeholder effort to improve information security by enabling the “good-faith testing, investigation, and/or correction of a security flaw or vulnerability” in computer programs.  As you review the public comments on this new petition, I urge you to consider expanding the good-faith security research definition to include both security and safety flaws or vulnerabilities, where safety includes bias and other harmful outputs. In considering this expansion, I urge the Copyright Office to continue to bind the exemption to research that is conducted in a safe environment, primarily to enhance the security or safety of computer programs, without facilitating copyright infringement.  Further, I encourage careful consideration of the exemption’s application to any research on technical measures that protect the authenticity or provenance of content from generative AI models.

              Sincerely,

###

WASHINGTON – Today, U.S. Sens. Mark R. Warner (D-VA), Chairman of the Senate Select Committee on Intelligence, and Thom Tillis (R-NC) – the bipartisan co-chairs of the Senate Cybersecurity Caucus – introduced the Secure Artificial Intelligence Act of 2024, legislation to improve the tracking and processing of security and safety incidents and risks associated with Artificial Intelligence (AI). Specifically, this legislation aims to improve information sharing between the federal government and private companies by updating cybersecurity reporting systems to better incorporate AI systems. The legislation would also create a voluntary database to record AI-related cybersecurity incidents including so-called “near miss” events.

As the development and use of AI grow, so does the potential for security and safety incidents that harm organizations and the public. Currently, efforts within the federal government – led by the National Institute of Standards and Technology (NIST) and the Cybersecurity and Infrastructure Security Agency (CISA) – play a crucial role in tracking of cybersecurity through their National Vulnerability Database (NVD) and the Common Vulnerabilities and Exposures Program (CVE), respectively. The National Security Agency (NSA), through the Cybersecurity Collaboration Center, also provides intel-driven cybersecurity guidance for emerging and chronic cybersecurity challenges through open, collaborative partnerships. However, these systems do not currently reflect the ways in which AI systems can differ dramatically from traditional software, including the ways in which exploits developed to subvert AI systems (a body of research often known as “adversarial machine learning” or “counter-AI”) often do not resemble conventional information security exploits. This legislation updates current standards for cyber incident reporting and information sharing at these organizations to include and better protect against the risks associated with AI. The legislation also establishes an Artificial Intelligence Security Center at the NSA to drive counter-AI research, provide an AI research test-bed to the private sector and academic researchers, develop guidance to prevent or mitigate counter-AI techniques, and promote secure AI adoption.

“As we continue to embrace all the opportunities that AI brings, it is imperative that we continue to safeguard against the threats posed by – and to -- this new technology, and information sharing between the federal government and the private sector plays a crucial role,” said Sen. Warner. “By ensuring that public-private communications remain open and up-to-date on current threats facing our industry, we are taking the necessary steps to safeguard against this new generation of threats facing our infrastructure.”

"Safeguarding organizations from cybersecurity risks involving AI requires collaboration and innovation from both the private and public sector,” said Sen. Tillis. "This commonsense legislation creates a voluntary database for reporting AI security and safety incidents and promotes best practices to mitigate AI risks. Additionally, this bill would establish a new Artificial Intelligence Security Center, within the NSA, tasked with promoting secure AI adoption as we continue to innovate and embrace new AI technologies."  

Specifically, the Secure Artificial Intelligence Act would:

·         Require NIST to update the NVD and require CISA to update the CVE program or develop a new process to track voluntary reports of AI security vulnerabilities;

·         Establish a public database to track voluntary reports of AI security and safety incidents;

·         Create a multi-stakeholder process that encourages the development and adoption of best practices that address supply chain risks associated with training and maintaining AI models; and

·         Establish an Artificial Intelligence Security Center at the NSA to provide an AI research test-bed to the private sector and academic researchers, develop guidance to prevent or mitigate counter-AI techniques, and promote secure AI adoption.

“IBM is proud to support the Secure AI Act that expands the current work of NIST, DHS, and NSA and addresses safety and security incidents in AI systems. We commend Senator Warner and Senator Tillis for building upon existing voluntary mechanisms to help harmonize efforts across the government. We urge Congress to ensure these mechanisms are adequately funded to track and manage today’s cyber vulnerabilities, including risks associated with AI,” said Christopher Padilla, Vice President, Government and Regulatory Affairs, IBM Corporation.

“Ensuring the safety and security of AI systems is paramount to facilitating public trust in the technology. ITI commends U.S. Senators Warner and Tillis for introducing the Secure Artificial Intelligence Act, which will advance AI security, encourage the use of voluntary standards to disclose vulnerabilities, and promote public-private collaboration on AI supply chain risk management. ITI also appreciates that this legislation establishes the National Security Agency’s AI Security Center and streamlines coordination with existing AI-focused entities,” said ITI President and CEO Jason Oxman.

“AI security is too big of a task for any one company to tackle alone,” said Jason Green-Lowe, Executive Director of the Center for AI Policy. “AI developers have much to learn from each other about how to keep their systems safe, and it's high time they started sharing that information. That's why the Center for AI Policy is pleased to see Congress coordinating a standard format and shared database for AI incident reporting. We firmly support Senator Warner and Tillis's new bill."

Full text of the legislation is available here. A one-page summary of the legislation is available here. 

###

WASHINGTON – Today, U.S. Sen. Mark R. Warner (D-VA) – Chairman of the Senate Select Committee on Intelligence, Co-Chair of the Senate Cybersecurity Caucus, and former technology entrepreneur – issued the statement below after President Joe Biden announced a new executive order on Artificial Intelligence.

“I am impressed by the breadth of this Executive Order – with sections devoted to increasing AI workforce inside and outside of government, federal procurement, and global engagement. I am also happy to see a number of sections that closely align with my efforts around AI safety and security and federal government’s use of AI. At the same time, many of these just scratch the surface – particularly in areas like health care and competition policy. Other areas overlap pending bipartisan legislation, such as the provision related to national security use of AI, which duplicates some of the work in the past two Intel Authorization Acts related to AI governance. While this is a good step forward, we need additional legislative measures, and I will continue to work diligently to ensure that we prioritize security, combat bias and harmful misuse, and responsibly roll out technologies.”