U.S. Department of Energy and GSA Advance Trustworthy AI and Scaled Deployment

U.S. Federal Agencies Accelerate Trustworthy AI Strategies

As artificial intelligence technology becomes increasingly prevalent across government agencies, ensuring the trustworthiness and security of AI systems — and efficiently deploying AI in large-scale real-world scenarios — has become a core challenge for governments worldwide. At a recent AI-focused conference, the U.S. Department of Energy (DOE) and the General Services Administration (GSA) each disclosed their strategic priorities regarding trustworthy AI and AI scaling, drawing widespread attention from the industry.

DOE: Mitigating Institutional Risk Through Trustworthy AI

The U.S. Department of Energy has explicitly designated "advancing trustworthy artificial intelligence and machine learning to reduce institutional risk" as a core priority. As the agency responsible for nuclear energy, energy infrastructure, and national laboratories, DOE faces AI risks of exceptionally high sensitivity.

In the energy sector, any deviation or error in AI systems could have severe consequences for critical infrastructure security. Accordingly, DOE emphasizes that trustworthiness principles must be embedded throughout the entire lifecycle of AI systems, including:

• Transparency: Ensuring AI decision-making processes are explainable and traceable
• Fairness: Eliminating data bias and algorithmic discrimination in models
• Security: Guarding against adversarial attacks and data breach risks
• Reliability: Guaranteeing stable operation of AI systems under extreme conditions

This strategic direction is highly consistent with the AI governance framework the U.S. government has been promoting in recent years and reflects the increasingly cautious approach federal agencies are taking when adopting AI technologies.

GSA: Focusing on Best Practices for Scaling AI Implementation

Unlike DOE's emphasis on risk governance, the General Services Administration (GSA) has placed its focus on "identifying best practices for scaling AI implementation." As the federal government's administrative services and procurement management agency, GSA is responsible for providing technology solutions to various departments, making its attention to AI scaling a cross-agency model of significance.

The core challenge GSA faces is: how to transform scattered AI pilot projects into replicable, sustainable large-scale applications? This involves challenges across multiple dimensions:

• Data Governance: Breaking down cross-agency data silos and establishing unified data standards
• Talent Development: Cultivating a civil service workforce with AI literacy
• Infrastructure: Building cloud computing and computational power platforms to support large-scale AI operations
• Procurement Mechanisms: Optimizing government procurement processes for AI technologies and services

Deeper Insights in the Context of Global AI Governance

The strategic deployments of these two major U.S. federal agencies reflect two critical issues in global AI governance: "trust" and "scale."

From a global perspective, the European Union has established a systematic AI regulatory framework through its AI Act, and China has also issued multiple AI governance standards and regulations on generative AI management. The practical explorations of U.S. federal agencies provide a reference model for how government departments can strike a balance between "innovation" and "risk management."

Notably, trustworthy AI and scaled deployment are not opposing concepts. Only by establishing robust trust mechanisms can AI technology truly gain the foundation for large-scale adoption; conversely, the experience and feedback accumulated through scaled applications will in turn drive continuous improvement of trustworthy AI standards.

Outlook: From Pilots to Systematic Deployment

As AI technology accelerates its journey from laboratories to real-world government operations, federal agencies are placing ever-greater emphasis on AI governance. The practices of DOE and GSA suggest that future government AI strategies will follow two major trends: first, deeply embedding trustworthiness requirements into every stage of AI system development and deployment; second, leveraging standardized and modular approaches to move AI from isolated experiments to systematic, scaled applications.

For governments around the world, how to unleash AI's enormous potential while safeguarding security and ethics remains an ongoing challenge. The explorations of U.S. federal agencies undoubtedly offer valuable practical pathways for this global endeavor.