Regulatory Consulting at Deloitte (GPS)

Overview

Deloitte operates the largest consulting arm in the world, and its Government & Public Services (GPS) practice is the most policy-relevant of the Big Four. GPS dominates federal technology delivery and has built an unusually developed responsible-AI apparatus around it: the Deloitte AI Institute for Government, the Deloitte Center for Government Insights (a research shop producing cutting-edge public-sector AI analysis), and a proprietary ‘Trustworthy AI’ framework that structures governance, risk, testing, and human oversight across the AI lifecycle. Deloitte’s scale makes it a bellwether for how responsible-AI norms actually land in government: it touches an enormous share of federal, state, and local AI modernization, from health and human services to defense and financial management. Its research is prolific and specific — analyses of how generative AI reshapes government work at the task level, guidance on scaling AI in state and local government, and convenings like the Federal Health AI Accelerator that stress-test trustworthy AI in high-stakes domains. Because Deloitte both advises on governance and delivers the systems, its Trustworthy AI framework functions less as marketing than as the de facto operating standard for a large slice of public-sector AI.

AI-Focused Regulatory Work

Deloitte GPS is where scale meets policy. Backed by the Deloitte AI Institute for Government and grounded in its Trustworthy AI framework, the practice has produced some of the most granular public-sector AI research in the market: a 2025 analysis of roughly 19,000 government tasks that maps which are genuinely amenable to generative AI (based on creative difficulty, accuracy needs, and context variability), and detailed work on scaling generative AI across state and local governments where, per NASCIO, more than half of state CIOs already report employees using GenAI tools. Deloitte’s Trustworthy AI framework operationalizes governance rather than gesturing at it — defining decision rights and ‘three lines’ roles, tiered risk minimums, standardized metrics for testing and incidents, and audit-ready evidence — and it has been extended into sensitive arenas through the Federal Health AI Accelerator, which red-teams AI for bias, discrimination, and privacy in federal health. The firm’s case work shows the payoff and the stakes together: a generative-AI categorization engine that enabled tens of millions in cost savings, alongside repeated insistence on human oversight. Deloitte’s people — Ed Van Buren, who leads AI strategic growth for GPS and directs the AI Institute for Government; researchers like Joe Mariani and Tasha Austin at the Center for Government Insights — give it both delivery muscle and a credible research voice. For a policy reader, Deloitte matters because the norms it encodes into ‘trustworthy AI’ propagate across a huge footprint of agencies, quietly setting the baseline for what responsible government AI looks like in practice.

Notable AI Policy Work

• Published ‘Generative AI and Government Work: An In-Depth Analysis of 19,000 Tasks’ (2025), a task-level framework for which government work is amenable to generative AI — among the most granular public-sector AI studies available.
• Operationalized a proprietary Trustworthy AI framework (governance, risk tiering, testing/red-teaming, audit-ready evidence, human oversight) across the AI lifecycle for government and enterprise clients.
• Runs the Deloitte AI Institute for Government and the Deloitte Center for Government Insights, producing sustained research on scaling AI in federal, state, and local government.
• Convenes the Federal Health AI Accelerator to build and red-team trustworthy AI for federal health (bias, discrimination, privacy, epidemiological modeling).
• Delivered public-sector GenAI case work — e.g., a generative-AI categorization/alerting engine that enabled ~$73M in cost savings — and enhanced state/local HHS benefits delivery with trustworthy GenAI.

Thought Leadership

1. Delivering on the Promise of AI in Government (Trustworthy AI)
   Explains what constitutes trustworthy AI in government and how public-sector leaders view generative AI versus their commercial counterparts.
2. Generative AI and Government Work: An Analysis of 19,000 Tasks (2025)
   A task-level study identifying three criteria for which government work generative AI can take on — and where occupations will feel adoption pressure.
3. Scaling Gen AI in US State and Local Governments (2025)
   Center for Government Insights research on the opportunities, challenges, and path to large-scale value for GenAI in state and local government.
4. AI Risk & Governance — Trustworthy AI framework
   Deloitte’s downloadable Trustworthy AI governance/risk framework: decision rights, tiered controls, standardized metrics, and audit-ready evidence.
5. Deloitte Federal Health AI Accelerator
   A convening to build and red-team trustworthy AI for federal health — targeting bias, discrimination, and privacy alongside safety and security.

Clients

Extensive federal, state, and local government clients — health and human services agencies, federal health, defense, and financial-management functions. Trustworthy-AI and GenAI delivery spans public benefits, epidemiological modeling, and citizen services.