OMB Memorandum on Artificial Intelligence Takes Two Steps Forward, but One Step Back

on

Washington, DC – UnidosUS, the nation’s largest Latino civil rights organization, welcomes the finalized agency guidance from the Office of Management and Budget (OMB) on Artificial Intelligence (AI). The memo sets out minimum practices for federal agencies to follow for uses of AI that impact safety or civil rights. Laura MacCleery, Senior Director of Policy for UnidosUS, issued the following statement:   

“The OMB’s final guidance provides important minimum steps for agencies to take in responsible use of AI across so many areas of government, including health, education, economic justice and immigration. It includes laudable calls to engage stakeholders, names specific ways that technologies may impact civil rights or safety and requires AI tools to work better than competing approaches—an appropriate check on hype around the systems. It also improves transparency and structures for AI governance across agencies. 

But the OMB memo sets a floor, not a ceiling. Unfortunately, it allows agency officials to decide whether an AI use can be excluded from even the minimum safeguards it offers, giving officials wide authority to exempt a system if they believe doing so would impede operations or increase overall risks to safety or rights. It also allows officials to sidestep them altogether if AI merely informed an action and was not the ‘principal basis’ for it. These large loopholes could easily swallow the policy, given internal pressures that AI officials will face. 

These policies will impact millions of Latinos as they interact with, and are affected by, government programs that will increasingly use AI tools. We know that AI models can be biased, inaccurate and used in ways that invade privacy, leading governments to overlook limitations and overstep their capabilities.  Instead of being encouraged to sidestep basic protections against abuse, agencies should be told to embrace opportunities to ensure systems are consistent, safe, effective, transparent, well-governed, privacy-enhanced and fair.  

Regardless of whether a policy requires it, federal agencies must ensure that AI uses align with our nation’s core civil rights, constitutional principles and democratic values. Communities of color, including Latinos, are among the first to be targets of tech surveillance and the last to benefit from its advances. Given its reach and power, to deploy AI responsibly and ethically requires new and innovative forms of governance. Checking a box — or escaping review altogether — will fail to address profound power imbalances at the heart of our shared technological future. For this reason, we will keep advocating for rules that more fully embrace the need for rigorous and transparent oversight in AI governance.” 

To learn more about UnidosUS’s commitment to including the Latino community in conversations around AI, see the documents below. 

  • Testimony on Three Pillars for AI work: Values, Voice, Investment 
  • Op-ed on Lessons Learned at the AI Insight Forum 
  • Written Statement of Janet Murguia at the first AI Insight Forum 
  • Press Statement on AI EO with Fact Sheet 
  • Comments and Fact Sheet on OMB Memorandum on AI Executive Order 
  • Comments on Federal Election Commission Rulemaking on Deepfakes in Elections 
  • Comments Addressing Concerns on Law Enforcement Technologies   
  • Written Testimony on Civil Rights Implications of Federal Use of Facial Recognition Technology