The Center for Democracy & Technology (CDT) has provided input to the Networking and Information Technology Research and Development National Coordination Office regarding priorities for a new AI Action Plan. This plan is mandated by Executive Order 14179, which aims to maintain and enhance the United States' global leadership in artificial intelligence (AI) while promoting human well-being, economic competitiveness, and national security.
CDT's submission highlights principles for trustworthy AI that have received bipartisan support. These principles were initially outlined during President Trump's first term through Executive Orders 13859 and 13960, as well as the Office of Management and Budget’s Memorandum M-21-06. They include evaluating risks to privacy, civil rights, civil liberties, safety; improving transparency and accountability; ensuring accuracy, reliability, effectiveness; and incorporating public input.
These principles have been integrated into the National Institute of Standards and Technology’s (NIST) AI Risk Management Framework. Congress has also endorsed similar principles through reports from the Bipartisan House Task Force on AI and the Bipartisan Senate AI Working Group. Companies have voluntarily adopted these principles in their AI governance commitments.
CDT recommends several elements for inclusion in the AI Action Plan:
1. **Continuing NIST's work**: NIST should continue developing voluntary standards with technical expertise to address risks to people's rights. The process should involve civil society expertise to identify community risks.
2. **Ensuring trustworthy AI in government**: The plan should promote safe, effective AI in government services with robust guardrails for safety, privacy, and civil liberties. It should focus on risk assessment, testing, centralized governance, privacy/security measures, public engagement, and transparency.
3. **Aligning national security use with civil liberties**: High-risk national security-related decisions using AI should be effectively reported to congressional committees with independent oversight mechanisms established.
4. **Advancing competitiveness through openness**: The plan should support open models to accelerate innovation while monitoring developments for potential risks without imposing premature export restrictions.
5. **Shaping responsible private sector use**: Agencies should help companies adopt governance measures ensuring their systems are effective without undermining rights or legal obligations.
The CDT emphasizes that both public and private sector development of AI should advance fundamental American values.
