Photo of Cynthia Cole

Cynthia Cole

Subscribe to Cynthia Cole's Posts
Contact:Read more about Cynthia Cole

On January 20, 2025, the first day of his second term, President Trump revoked Executive Order 14110 on Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence (the “Biden Order”), signed by President Biden in October 2023. In doing so, President Trump fulfilled a campaign pledge to roll back the Biden Order, which the 2024 Republican platform described as a “dangerous” measure. Then on January 23, 2025, President Trump issued his own Executive Order on AI, entitled Removing Barriers to American Leadership in Artificial Intelligence (the “Trump Order”). Here, we examine some of the practical implications of the repeal and replacement of executive orders by Trump and what it means for businesses.

Overview of the Executive Orders

Building on the White House’s 2022 Blueprint for an AI Bill of Rights, the Biden Order outlined a sweeping vision for the future of AI within the federal government, including seven high-level objectives: (1) Ensuring the Safety and Security of AI Technology; (2) Promoting Innovation and Competition; (3) Supporting Workers; (4) Advancing Equity and Civil Rights.; (4) Protecting Consumers, Patients, Passengers, and Students; (5) Protecting Privacy; (6) Advancing Federal Government Use of AI; and (7) Strengthening American Leadership Abroad.

The Biden Order directed various measures across the federal apparatus –imposing 150 distinct requirements on more than 50 federal agencies and other government entities, representing a genuinely whole-of-government response.

Although the bulk of the Biden Order is addressed to federal agencies, some of its provisions had potentially significant impacts on private sector entities. For example, the Biden Order directed the Commerce Department to require developers to report on the development of higher risk AI systems.  Similarly, the Biden order directed the Commerce Department to establish requirements for domestic Infrastructure as a Service (IaaS) providers to report to the government whenever they contract with foreign parties for the training of large AI models. The Biden Order also open-endedly instructed federal agencies to use existing consumer protection laws to enforce against fraud, unintended bias, discrimination, infringements on privacy, and other harms from AI—a directive various federal regulators actioned under the Biden administration.

Other than the definition of AI, the Trump Order and Biden Order share no similarities (both Orders point to the AI definition from 15 U.S.C. 9401(3), namely: “a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments”). The Trump Order does not contain specific directives (such as those in the Biden Order), but instead articulates the national AI policy to “sustain and enhance America’s global AI dominance in order to promote human flourishing, economic competitiveness, and national security.” The Trump Order directs a few specific roles within the administration to develop an Artificial Intelligence Action Plan within 180 days (i.e., by July 22, 2025) to achieve the policy objective articulated in the Trump Order. The Trump Order directs these same roles within the administration to review the policies, directives, regulations, orders, and other actions taken pursuant to the Biden Order and to suspend, revise, or rescind any such actions that are inconsistent with the Trump Order’s stated policy. In cases where suspension, revision, or rescission of the prior action cannot be finalized immediately, the heads of agencies are instructed to “to provide all available exemptions” in the interim.

Practical Impacts

The practical effect of the revocation of the Biden Order—and the options available under the Trump Order—will vary depending on the measure. Although there are widespread impacts from the revocation of the Biden Order’s mandates across multiple initiatives and institutions, below are those that are expected to have a significant impact on private sector entities engaged in the development or use or AI.

Reporting requirement for powerful AI models: As noted, the Biden Order directed the Department of Commerce to establish a requirement for developers to provide reports on “dual-use foundation models” (broadly, models that exhibit high levels performance at tasks that pose a serious risk to security, national economic security, national public health or safety). Pursuant to the Biden Order, the Bureau of Industry and Security’s (BIS), a Commerce Department agency, published a proposed rule to establish reporting requirements on the development of advanced AI models and computing clusters under its Defense Production Act authority, but had not issued a final rule prior to the revocation of the Biden Order. It is likely that the new administration will closely scrutinize this reporting requirement and may take action to block the adoption of the final rule if it is found to be inconsistent with the policy statement in the Trump order.Continue Reading AI Tug-of-War: Trump Pulls Back Biden’s AI Plans

On May 17, 2024 Colorado Governor Polis signed the landmark Colorado AI Act (Senate Bill 24-205) into law. Colorado is now the first US state with comprehensive AI regulation, adopting a classification system like the European Union’s recent AI Act. The law will take effect February 1, 2026. 

The law exempts small employers (fewer than fifty full-time employees) from some of its requirements but otherwise requires companies to take extensive measures to protect Colorado residents against harms such as algorithmic discrimination.

SB 205’s Details

SB 205 requires “developers” and “deployers” of “high-risk artificial intelligence systems” to use “reasonable care” to protect Colorado resident consumers from any known or reasonably foreseeable risks of “algorithmic discrimination.” As written, the law most likely applies to both creators of high-risk AI systems, as well as employers adopting high-risk AI technologies within their organization.  Continue Reading From Brussels to Boulder: Colorado Enacts Comprehensive AI Law with Significant Obligations for Employers on the Heels of the EU AI Act

Baker McKenzie is pleased to invite you to an afternoon exploring strategy and risk in the year of the dragon.

Alongside industry leaders from Meta and Dayforce, this will be an interactive discussion exploring how businesses can harness the power and auspiciousness of the mythical dragon in building a comprehensive data, AI and cyber…

Special thanks to co-presenters Teresa Michaud and Bradford Newman.

California’s CLE Compliance Deadline Is Approaching…    
We can help!

If your last name starts with H-M, you are probably well aware that your CLE compliance deadline is right around the corner – February 1, 2024. In addition to the general credit requirement, the state of California requires all attorneys to complete:

  • At least four hours of legal ethics
  • At least one hour on competence issues
  • At least two hours on the elimination of bias in the legal profession and society. Of the two hours, at least one hour must focus on implicit bias and the promotion of bias-reducing strategies.

Our lawyers will offer three virtual sessions, focused on key considerations for AI development and utilization, to help you meet your CLE requirements. These sessions will also offer CLE credit in the states of Illinois, Texas, and New York. Participants requesting CLE for other states will receive uniform CLE certificates. 

Please register and let us know which individual session(s) you plan to attend. We look forward to your participation!

Promoting Unity: Overcoming the Risks of Bias and Prejudice in the Workplace

Tuesday, January 16, 2024 | 1:00 – 2:00 pm Pacific
1 hour Elimination of Bias credit (pending approval)Continue Reading California AI CLE Series

In Raines v. U.S. Healthworks Medical Group, the California Supreme Court expanded the definition of an “employer” under the state’s discrimination statute to include certain third-party business entities that perform employment-related functions on behalf of employers. These agents may now be deemed “employers” such that they can be directly liable for employment discrimination under the Fair Employment and Housing Act for certain activities that they carry out on behalf of employers.

Overview of Raines

The Raines‘ plaintiffs were job applicants who received offers of employment that were conditioned on the successful completion of pre-employment medical screenings conducted by a third-party company that used automated decision-making. Plaintiffs alleged that the screening form contained intrusive questions regarding their medical history that violated FEHA. They brought claims against their employers, as well as the third-party provider that conducted the medical screening. The question for the Court was whether business entities acting as agents of an employer, can be considered “employers” under FEHA and held directly liable for FEHA violations caused by their actions.

The Court examined the plain language in FEHA’s definition of “employer” and concluded that the definition did indeed encompass third-party corporate agents like the medical provider in his case. FEHA defines an employer as “any person regularly employing five or more persons, or any person acting as an agent of an employer, directly or indirectly.” Here, the Court reasoned, recognizing the medical provider as an agent of the employer extended liability to the company most directly responsible for the FEHA violation.Continue Reading Automated Decision-Making and AI: California Expands FEHA Liability to Include Third-Party Business Agents of Employers

New York may soon restrict employers and employment agencies from using fully-automated decision making tools to screen job candidates or make other employment decisions that impact the compensation, benefits, work schedule, performance evaluations, or other terms of employment of employees or independent contractors. Draft Senate Bill 7623, introduced August 4, aims to limit the use of such tools and requires human oversight of certain final decisions regarding hiring, promotion, termination, disciplinary, or compensation decisions. Senate Bill 7623 also significantly regulates the use of certain workplace monitoring technologies, going beyond the notice requirements for workplace monitoring operative in New York since May 2022 and introducing data minimization and proportionality requirements that are becoming increasingly common in US state privacy laws.

While there is not yet a federal law focused on AI (the Biden administration and federal agencies have issued guidance documents on AI use and are actively studying the issue), a number of cities and states have introduced bills or resolutions relating to AI in the workplace. These state and local efforts are all at different stages of the legislative process, with some paving the path for others. For example, New York City’s Local Law 144 took effect on July 5, prohibiting employers and employment agencies from using certain automated employment decision tools unless the tools have undergone a bias audit within one year of the use of the tools, information about the bias audit is publicly available, and certain notices have been provided to employees or job candidates (read more here).

If enacted, Senate Bill 7623 would take things much further. Here are some of the most significant implications of the draft legislation:Continue Reading Check Yourself Before You Wreck Yourself: New York and Other States Have Big Plans For Employer Use of AI and Other Workplace Monitoring Tools