The White House recently issued two executive orders shifting the administration’s strategy regarding AI development by promoting federal oversight for new AI models and deploying select cutting-edge AI technologies to aid the warfront.
These orders signal a dramatic change to the administration’s hands-off approach to AI development, which to this point largely supported limited regulations to spur innovation. However, as cybersecurity risks escalate, the government is expanding its influence over what and how select AI models are implemented to address unique risks and close security gaps created by these technologies.
The June 2 executive order most notably encourages AI developers to voluntarily give the federal government oversight of new AI models before their public release. While compliance isn’t required, the order represents the administration’s biggest shift toward AI regulation, going against its permissive stance toward AI development since its return to office.
Over the next 60 days, various federal branches as directed by the order will create a benchmarking process to assess the capabilities of AI models and determine the criteria needed for these technologies to be considered a “covered frontier model” – representing the most advanced version of AI technology available. They will also design the voluntary framework enabling AI developers to:
- Work with the government to receive that designation.
- Give the government a 30-day period to evaluate the model before public release.
- Determine what partners will have early access to the model.
These steps will help accelerate frontier model deployment while assuring they are safe and usable, posing minimal to no risk to national security. The order also stipulates that the process does not require a mandatory license, clearance or permitting requirement for these technologies.
The order includes measures carried out over the next 30 days to further bolster the nation’s cyber defense across National Security Systems, the Department of War and release Binding Operational Directives to guide AI development and implementation. These directives include:
- Streamlining and prioritizing the cyber defense of information systems.
- Creating or expanding programs and cybersecurity services that improve AI-enabled defensive tools.
- Providing access to tools and services that include covered frontier models for agencies, authorities and critical infrastructure operators.
Select federal agencies have also been directed to create an AI cybersecurity clearinghouse over a 30-day period. The clearinghouse will review security vulnerabilities discovered by AI models and accelerate remediation packages. The Office of Management and Budget (OMB) will examine existing funding sources that can be redirected to facilitate efforts to create AI vulnerability detection solutions.
While much of the order is voluntary, its release has prompted immediate reaction from the industry as AI developers work to determine how these measures will influence development and deployment in both the public and federal sectors.
A few days after the first order was released, the White House unveiled a second directive specifying cybersecurity measures and cutting-edge AI deployment for the nation’s armed forces. The “Memorandum on AI in the National Security Enterprise” establishes a framework to ensure warfighters and intelligence professionals have access to the most advanced and security AI systems available.
Measures targeted by the memorandum include:
- Accelerating AI adoption and adapt commercial and open-source technologies for mission use.
- Onboarding advanced AI models from multiple vendors and building out next-generation, high-security computing facilities to strengthen national security capabilities.
- Establishing an AI National Security Strategic Reserve composed of non-governmental experts.
- Releasing an updated directive on autonomy in weapon systems and annually reviewing key guidance to stay atop the AI frontier.
- Ensuring no entity can disable, degrade of modify critical AI systems without approval.
- Offering partnerships with private-sector companies to advance cutting-edge AI technology.
- Rescinding and replacing an AI directive established by the previous administration.
Photo by Sergei Starostin from Pexels
For more of the latest from the expansive government marketplace, check Government Market News daily for new stories, insights and profiles from public sector professionals. Check out our national contracting newsletter here.
