How the AI Action Plan reshapes federal data strategy

In July, the Trump Administration released its sweeping AI Action Plan, a bold roadmap designed to secure American leadership in artificial intelligence through innovation, infrastructure, and international engagement. With over 90 federal policy actions, the Winning the Race plan emphasizes deregulation, open-source development, and AI systems that reflect the country's values while protecting national interests. Cohesity supports this approach and recognizes it as an important step in advancing responsible AI innovation.

The plan signals a decisive shift in how federal agencies approach data management, security, and technological sovereignty. For IT leaders, the takeaway is clear: Data is no longer a byproduct of operations. It is a strategic asset that will determine mission success in the AI era.  

World-class datasets are a national security imperative

The AI Action Plan elevates data quality from an operational concern to a national objective. The document states, “High-quality data has become a national strategic asset as governments pursue AI innovation goals and capitalize on the technology’s economic benefits. The United States must lead the creation of the world’s largest and highest quality AI-ready scientific datasets.”

That doesn’t mean agencies need to collect more data. It stresses that the data being collected needs to be able to power AI systems effectively. The Plan calls for “minimum data quality standards for the use of biological, materials science, chemical, physical, and other scientific data modalities in AI model training” and directs agencies to “establish secure compute environments within NSF and DOE to enable secure AI use-cases for controlled access to restricted Federal data.”

Federal agencies sit on massive data repositories. Federal open-data holdings now exceed 312,000 datasets across 100+ organizations. Yet only 18% of federal data leaders are confident their current data-management systems can support AI initiatives, even as OMB M-24-10 directs agencies to ensure “output of any generative-AI system is traceable to authoritative data.”

Most agencies already have the raw material but lack the infrastructure to transform it into AI-ready datasets. The challenge is data accessibility and quality assurance at scale.

Security-first architecture

The AI Action Plan treats security as fundamental to AI success. The Plan establishes “an AI Information Sharing and Analysis Center (AI-ISAC), led by DHS, in collaboration with CAISI at DOC and the Office of the National Cyber Director, to promote the sharing of AI-security threat information and intelligence across U.S. critical infrastructure sectors.” 

More significantly, it calls for “new technical standards for high-security AI data centers, led by DoD, the IC, NSC, and NIST at DOC” and mandates agencies to “promote Secure-By-Design AI Technologies and Applications.” 

This security-first approach recognizes that AI systems are only as trustworthy as the data that powers them. The NSA and CISA note data poisoning as a top AI risk, and academic testing shows that poisoned data can cut model accuracy by up to 27%. With that in mind, agencies need control over every part of the AI ecosystem, from data integrity to information architecture.

The Plan’s emphasis on “rigorous evaluations as a critical tool in defining and measuring AI reliability and performance in regulated industries” signals that agencies need more than basic cybersecurity. They need systems that can prove data provenance and maintain audit trails throughout the AI lifecycle.

Breaking free with open-weight models

Perhaps the most strategically significant aspect of the AI Action Plan is its explicit backing of open-weight AI models. The Plan states unequivocally that “open-source and open-weight AI models are made freely available by developers for anyone in the world to download and modify. Models distributed this way have unique value for innovation because startups can use them flexibly without being dependent on a closed model provider.” 

For federal agencies, this represents a fundamental shift. The Plan recognizes that “many businesses and governments have sensitive data that they cannot send to closed model vendors.” It emphasizes that open models are “essential for academic research, which often relies on access to the weights and training data of a model to perform scientifically rigorous experiments.” 

The strategic implications are profound. By prioritizing open-weight models, agencies can maintain control over their data while avoiding dependence on proprietary APIs that could change terms, pricing, or availability without notice. With these changes, the Federal Government has the potential to be at the forefront of AI-powered innovation.

The infrastructure reality check

Theory has to meet operational reality. The AI Action Plan’s ambitious goals would require agencies to be able to surface petabytes of versioned, immutable data on demand with push-button simplicity for data readiness. Most agencies aren’t close to this capability.

The data agencies need for AI already exists. It’s sitting in backup systems, email archives, document repositories, and file shares across hybrid cloud environments. The challenge isn’t creating new data. Agencies have to make existing data discoverable, searchable, and usable for AI applications while maintaining security and compliance standards. To do that, agencies need solutions that can transform their existing data infrastructure for use in AI-ready platforms.

Cohesity’s role in helping agencies be AI-ready

The AI Action Plan’s priorities align directly with the Cohesity Data Cloud and Cohesity Gaia platforms to meet these exact requirements:

Data readiness at scale: Gaia can transform existing backup repositories into searchable, AI-ready datasets without requiring agencies to build separate data lakes or compromise on data governance. Agencies need solutions that can process emails, PDFs, and documents—exactly what Gaia specializes in indexing and surfacing.

Security and governance: With granular role-based access controls and immutable backup architecture, the Cohesity platform helps agencies maintain the data integrity and audit trails that the AI Action Plan emphasizes. The platform’s integration with leading security alliances provides the ecosystem approach to threat detection that the Plan’s AI-ISAC envisions.

Open model integration: Gaia’s retrieval-augmented generation architecture works with multiple LLMs—currently OpenAI GPT models and Google Vertex, with more to come. Agencies need the flexibility to choose models that align with their security requirements and mission. This multi-model approach supports the Plan’s emphasis on avoiding vendor lock-in.

Looking ahead

The AI Action Plan is a roadmap for maintaining technological leadership in an increasingly competitive global landscape. Agencies that can quickly operationalize high-quality, secure datasets while leveraging open-weight models will have significant advantages in delivering citizen services and accomplishing their missions. As more than 70% of agencies will use AI to enhance administrative decision-making by 2026, the infrastructure decisions agencies make today will determine their competitive position for the next decade.

Delivering on the AI Action Plan means moving from pilots to production systems that can handle mission-scale data volumes, security requirements, and evolving AI technologies. The framework is set—now agencies must build the infrastructure to make it real.

At Cohesity, we are ready to help you explore how your agency can implement AI-ready data management. Learn more about Cohesity’s approach to conversational AI and secure data platforms, here.