> ## Content Index > Fetch the complete content index at: https://aiadoption.org/llms.txt > Use this file to discover other available public pages before exploring further. # Industry now dominates US congressional AI hearings - URL: https://aiadoption.org/ai-analysis/industry-now-dominates-us-congressional-ai-hearings/ - Published: 2026-09-23T00:50:38.000Z - Updated: 2026-09-23T00:50:38.000Z - Description: Industry witnesses now make up 37% of US congressional AI hearings, up from 13% in 2017. Government's share fell from 35% to 10%. Academia from 26% to 15%. The voice that defines AI policy in Washington has shifted decisively toward industry over five Congresses. - Author: Jassie - Tags: AI Analysis, AI Leadership, AI Policy, AI Strategy If your model of how US AI policy gets shaped at the federal level assumes a balanced witness mix across academia, government, industry, and civil society, the 2017-2025 congressional hearing data shows the balance has shifted decisively toward industry. Industry's share of US congressional AI witnesses rose from 13% in the 115th Congress (2017-2018) to 37% in the 119th Congress (2025), making it the largest single witness group. Over the same period, government's share fell from 35% to just 10%. Academia from 26% to just 15%. Civil society and nonprofit ("other") grew from 26% to a 38% share. The voice that defines AI policy at the federal hearing level has shifted decisively. Witness composition in US congressional AI hearings by sector, 115th–119th Congress: The volume context: total witness numbers across hearings have grown almost twentyfold since 2017, from 5 to 102 in 2025\. The growth accelerated after 2022, from 18 witnesses in 2022 to 131 in 2023, with the 2025 count of 102 representing sustained engagement at the post-ChatGPT level. Congressional attention to AI is now consistently high. The composition of that attention (who Congress is hearing from) has shifted in measurable ways. Witnesses in US congressional AI hearings by year, 2017–25: ## **The industry-share growth from 13% to a 37% share is the leading dynamic. Three structural drivers explain the shift.** **The first driver: industry's expanding role in overall AI development.** Private companies now account for the majority of frontier model releases and infrastructure investment. The companies operating at frontier scale are positioned as both more relevant sources of technical input on AI policy questions and as more active participants in shaping the policy environment in which they operate. The witness-share shift partly reflects an underlying shift in where AI capability is concentrated. **The second driver: the relative decline of government and academic AI capability has reduced their relative voice in policy formation.** Government's share fell from 35% to just 10% over the same period. Academia from 26% to just 15%. As industrial labs scaled, the technical authority on frontier AI questions concentrated in them, and the government and academic voices that historically dominated technical policy discussions became relatively less central. The 2017 hearing where a government technical witness explained how a class of AI systems works is a different conversation than the 2025 hearing where an industrial lab CTO explains the same class of systems, and the witness mix reflects which voice is most authoritative on the current frontier. **The third driver: civil society and nonprofit voice has grown from 26% to a 38% share, reflecting the rise of organisations specifically focused on AI policy (think tanks, advocacy organisations, public interest technology groups).** The "other" category contains both AI-focused civil society organisations and broader policy organisations that have added AI to their portfolio. The category growth is real but heterogeneous. The subject-area breakdown adds context. Between 2017 and 2025, the most-discussed subjects in AI-related congressional hearings were: - General AI governance: 113 total witnesses - National security and defence: 74 - Finance and economic policy: 39 - R&D and innovation: 39 - Government modernisation: 38 - User risks, rights, and protections: 35 - Privacy and legal: 33 - Education and workforce: 28 - Health and biomedical AI: 18 (House 9, Senate 9, the only area with chamber parity) - Industry and commerce: 20 The general AI governance category dominates. The cross-cutting topics that affect many sectors (governance, national security, R&D) generate more witness activity than vertical sector deep-dives. The implication: organisations whose AI strategy intersects with these horizontal topics are more likely to be exposed to federal policy formation through the hearing pipeline than organisations whose AI work is vertical-specific. The data observation: the witness-share shift is a structural feature of US AI policy formation rather than a short-term aberration. The drivers (concentration of frontier capability in industry, relative decline of government and academic technical authority, growth of AI-focused civil society) are durable. The trajectory through 2026-2028 suggests continued industry prominence, possibly stabilising in the 35–45% range, with the other three groups (government, academia, civil society) competing for the remaining share. ## Three implications for organisations whose AI strategy intersects with federal policy formation follow. **The first implication: federal AI policy formation now reflects the views, interests, and framings of frontier-AI industry participants more than it did a decade ago.** Organisations whose interests align with the frontier industry have stronger relative voice; organisations whose interests diverge from frontier industry (smaller AI players, downstream deployers, sectoral specialists) have weaker relative voice and may need to invest more deliberately in their policy engagement infrastructure to be heard. **The second implication: the "general AI governance" category, which dominates the witness data, is where most federal policy gets shaped.** The horizontal-policy framing is the venue. Vertical-sector engagement matters but is smaller in volume. Organisations whose primary federal policy interest is sectoral (healthcare, finance, manufacturing) may need to engage in horizontal-policy venues alongside their sector-specific advocacy to influence the underlying framework. **The third implication: the academic and civil society voice is now smaller in relative share but still present.** Organisations seeking to shape AI policy through evidence-based or rights-based framing have access to channels (academic partnerships, civil society coalitions) that carry meaningful weight even at reduced share. The strategic move is to engage these channels deliberately rather than assume the policy environment will absorb evidence-based framings independently. > The scale of the shift: the federal AI policy voice mix has moved by 24 percentage points (industry up) and 25 percentage points (government down) over five Congresses. The shift is large enough to materially change which framings, interests, and technical perspectives dominate federal AI policy formation. The 2026-2028 federal AI policy environment will continue to reflect this composition. Organisations whose strategic interests depend on the federal policy formation process should plan around the actual witness mix rather than the historical balance that the 2017 data captured. For organisations setting federal AI policy engagement strategy, the planning anchor should engage with the witness composition directly. The composition is publicly visible. The trajectory is measurable. The implications for whose framings carry weight in hearing rooms are direct. Plans that incorporate this picture will be better calibrated than plans that assume a balanced witness mix that the data shows no longer exists. --- ### Sources - **Primary**: Stanford AI Index 2026, Chapter 8 (Policy and Governance) 8.4 — [hai.stanford.edu/ai-index/2026](http://hai.stanford.edu/ai-index/2026?ref=aiadoption.org) - **Hearing data**: AI Index 2026 US congressional AI-related hearings tracker — 115th–119th Congress, witness composition by sector and subject area - **Cross-reference**: Stanford AI Index 2026 Chapter 1 (R&D) and Chapter 4 (Economy) — industry concentration of frontier capability and investment