Cantwell Outlines Comprehensive Governance Framework for Safe and Secure Frontier AI

WASHINGTON, D.C. – U.S. Senator Maria Cantwell (D-Wash.), Ranking Member of the Senate Committee on Commerce, Science and Transportation, outlined a comprehensive AI governance framework built upon six principles that must drive the legislative and executive actions needed to address catastrophic risks and other AI harms.

“To manage risks from advanced AI systems we need clear safety standards, continuous testing, and reporting of serious failures,” Sen. Cantwell said. “America can lead the world in AI by building systems that are not only more capable, but safer and more secure.”

Frontier AI systems require constant oversight and vigilance at home and abroad. The United States should establish clear, transparent, enforceable safety standards, developed by federal experts at NIST in coordination with other relevant federal agencies. Covered models should not be released until they have undergone an independent audit confirming that they meet these safeguards and standards.

Sen. Cantwell’s framework is built around six principles, many of which she has promoted for years as Ranking Member and former Chair of the Commerce Committee:

Clear, Enforceable Federal Safety Standards for Frontier AI Development and Deployment
Continuous Testing, Reviewing and Auditing to Verify Standards are Met
Transparency and Accountability Through Disclosure, Oversight and Consequences
Public Private Partnerships to Benefit the Public Good
Protections for Children, Support for Workers and Human Oversight of Consequential AI
U.S. Led Global Cooperation in Adopting and Upholding Common AI Safety and Security Standards

ENFORCED STANDARDS

Set standards for safe AI development and deployment

Federal Standards: NIST should develop clear, transparent, measurable, risk-based standards for AI systems that could cause catastrophic harm
Catastrophic Risks: Standards must address catastrophic risks, including AI-enabled cyberattacks; chemical, biological, radiological and nuclear threats; potential for loss of human control; and automated improvement and autonomous agents escaping secure testing environments
Risk Management and Controls: Standards should establish proportionate safeguards and risk-management requirements, with heightened review and oversight for systems that exhibit dangerous capabilities or are capable of self-improvement that may bypass safety controls
Open-Source Protections: Specialized standards should be developed for open-source AI models to reduce the risks of catastrophic misuse while providing developers with the clear guidance necessary to accelerate U.S. leadership of the open-source ecosystem
Keeping Pace: Federal standards should be regularly updated as AI evolves

INDEPENDENT TESTING

Verify AI safety through continual review, constant oversight

Continuous Testing: Government and independent experts should provide constant oversight and continually scrutinize and stress test covered AI models against catastrophic risk standards, particularly where testing requires specialized expertise, sensitive information or secure facilities
Built-In Safeguards: AI systems must be stress-tested to ensure they halt self-modification that bypasses safety controls, monitoring or human oversight
Independent Audits: Audit firms should conduct comprehensive audits of AI developers, akin to outside auditors examining a company’s financial statements, to ensure AI products are meeting applicable federal standards and safeguards

TRANSPARENCY AND ACCOUNTABILITY

Safety requires disclosure, oversight, and consequences

Independent Access: Developers should provide qualified independent testers and auditors with access and information necessary to assess catastrophic risks, investigate serious incidents and verify that safety and security safeguards are functioning as intended
Public Disclosure: Developers should provide clear, plain-language disclosures of material safety and security risks, and mitigation modeled on the risk disclosures public companies provide to investors
Incident Reporting: Developers must promptly report material AI safety and security incidents, including loss of control over autonomous agents, dangerous cyber capabilities, failures of critical safeguards and unsafe recursive self-improvement
Accountability: Developers must remain liable under applicable civil and criminal law for foreseeable harms and unlawful conduct arising from their AI systems, including when their design, deployment, modification or failure to implement reasonable safeguards materially contributes to that harm
Whistleblower Protection: Employees and contractors who report safety, security or compliance concerns should be protected from retaliation

PUBLIC PRIVATE PARTNERSHIPS

Get the best experts working for the public good

Defensive AI: Government should work with industry to develop and deploy defensive AI that detects and contains malicious activity, identifies and patches cyber vulnerabilities and protects critical infrastructure from increasingly automated attacks
Shared Capabilities: Leading AI companies should contribute compute, technical expertise and advanced defensive tools, so small and medium sized businesses, critical infrastructure operators and public institutions can benefit from capabilities they may not be able to build on their own
Public-Interest Compute: Government and industry should expand access to advanced computing and AI tools for universities, researchers and public-interest projects working on national security, scientific discovery, health, energy and other national priorities
Local Resilience: Federal agencies should equip state and local governments and private-sector operators with technical assistance, training and resources needed to protect critical infrastructure and essential services

DEFEND AGAINST SOCIAL HARM

Make AI work for people, not against them

Online Child Safety: AI products used by children should protect their safety and healthy development, rather than exacerbate social media harms such as addictive design, harmful content or data exploitation
Worker Opportunity and Protections: AI companies should help fund apprenticeships, education and other worker training so that AI expands workers’ skills, productivity and earning power rather than displacing them, while guardrails protect the products of human creativity and incentivize employers to use AI to improve jobs rather than eliminate them
Human Judgment: When AI informs consequential decisions about employment, health care, credit or other essential services, it must be transparent and fair and be subject to meaningful human oversight—particularly where AI-assisted prior authorization, such as CMS’s WISeR model, could delay or deny needed care
Recourse: People should be able to understand when and how an AI system materially affected a consequential decision, challenge the outcome and receive timely review by a qualified human decision-maker

ADVANCE GLOBAL COOPERATION

Lead global adoption of trusted American AI

Align Standards: The United States should work with allies and partners to adopt common, rigorous AI safety and security standards that reduce conflicting requirements and expand global markets for trusted American AI
Standards Leadership: The United States should lead in international standards-setting bodies, including ISO/IEC, to translate U.S. technical leadership in AI testing, risk measurement and safeguards into widely adopted global standards
Shared Risk Information: Allies should establish secure mechanisms to share information on serious AI incidents, emerging catastrophic capabilities, evaluation methods and effective safeguards
U.S.–China Crisis Channel: The U.S. should establish a secure channel with China like the Soviet era red phone for rapid communication when a significant AI incident or system-control failure could threaten either country or international security
Secure Access: Export controls should ensure that America’s most advanced AI technologies are not made accessible to our adversaries and available only to countries that protect them from diversion and misuse and uphold U.S.-led safety and security standards

Additional information about Sen. Cantwell’s leadership on AI policy and legislation can be found HERE.

###

Legal Disclaimer:

EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.