News that Caught Our Eye #77

Published on September 25, 2025

Summary

Congress will roll out Microsoft Copilot for 6,000 House staffers. Ohio launches an AI-powered public safety app while Albania’s AI “minister” echoes a South Korean experiment. A Supreme Court ruling in Spain affirms the public’s right to algorithmic transparency, and a Times investigation reveals that Trump promised advanced U.S. AI chips to the UAE—even amid fears they could reach China—as $2 billion flowed into a Trump-linked crypto firm. Meanwhile, a new People Powered guide reminds policymakers that choosing a participation platform is a structural decision, not a technical one. This and more in this week’s News that Caught Our Eye.

Upcoming InnovateUS Workshops

InnovateUS delivers no-cost, at-your-own-pace, and live learning on data, digital, innovation, and AI skills. Designed for civic and public sector, programs are free and open to all.

Coming up this week:

AI and Policing: AI Policy Development, September 25, 2025, 2:00 PM ET
Designing Smarter Engagement with AI, September 30, 2025, 3:00 PM ET
AI in Police Operations, October 1, 2025, 2:00 PM ET

AI for Governance

House Staffers to Use Microsoft Copilot for Constituent Services and Workflow Efficiency

Miranda Nazzaro on September 17, 2025 in FedScoop

House Speaker Mike Johnson announced that 6,000 House staffers will gain access to Microsoft Copilot starting this fall. The deployment, framed as a pilot to modernize workflows and constituent service, comes after a prior ban on the tool and follows two years of internal reviews by the House AI Working Group. The Chief Administrative Officer confirmed a deal to bring Microsoft 365 tools, including Copilot, to Congress. The initiative reflects growing momentum across federal agencies to experiment with generative AI, with several companies offering steeply discounted tools as public-sector demand rises.

Read article

AI for Governance

Improving How State and Local Governments Use AI

Sara Frueh on September 15, 2025 in National Academies

A new National Academies rapid expert consultation urges state and local governments to take a measured, problem-driven approach to AI adoption, eschewing hype in favor of human-centered design and real-world evaluation. Experts Suresh Venkatasubramanian (Brown), Nathan McNeese (Clemson), and Leila Doty (San José) stress the importance of pilot testing, domain expert involvement, and partnerships with universities and civil society. Drawing lessons from failures like Michigan’s flawed unemployment fraud algorithm, the report outlines practical strategies for experimentation, governance, and feedback loops. The piece also highlights the growing GovAI Coalition, a peer-driven network of over 850 agencies working to improve AI procurement and transparency.

Read article

AI for Governance

Strengthening Democracy: The Capacity of AI-Powered Insights for Enhancing Policy Deliberation and Transparency

Nibin Koshy and Greig Mordue on September 17, 2025 in Journal of eDemocracy and Open Government

This article explores how AI-powered insights can strengthen democratic institutions by enhancing transparency and improving the quality of policy deliberation. Koshy and Mordue argue that AI tools, when governed responsibly, can expand access to diverse datasets, surface latent public needs, and support evidence-based decisions that are more inclusive. Drawing on case studies and a structured literature review, the authors identify key enablers such as data interoperability, human-centered design, and algorithmic accountability. While the risks of opacity and exclusion remain real, the piece makes a compelling case that democratic resilience can be bolstered when AI is used not just to automate tasks, but to broaden who participates in governance and how.

Read article

AI for Governance

Don't Treat Albanian's AI Minister Like a Stunt

Beth Simone Noveck on September 23, 2025 in Washington Post

Albania’s appointment of an AI chatbot named Diella as a “digital minister” sparked ridicule in U.S. media, but that reaction misses the point. In her WaPo letter and accompanying blog post, Beth Simone Noveck explains how Seoul showed 25 years ago that transparency and accountability built into digital systems can blunt corruption. Albania’s chatbot minister will stand or fall on the same test: whether it reduces opportunities for bribery.

Read article

Governing AI

Spain’s Supreme Court rules in favor of algorithmic transparency

Editorial Staff on September 18, 2025 in Civio

In a landmark ruling, Spain’s Supreme Court ordered the government to release the source code of BOSCO, the algorithm that determines eligibility for energy subsidies. The decision affirms that public-sector algorithms must be subject to transparency and oversight, rejecting national security and intellectual property as blanket exceptions. The Court framed access to algorithmic decision-making as a constitutional right linked to democratic accountability and citizen participation. Civio’s seven-year legal battle sets a powerful precedent for digital democracy and the public’s right to know how automated systems impact their lives.

Read article

AI and International Relations

LISTEN: The U.A.E. Got A.I. Chips. Trump’s Inner Circle Got Crypto Riches. A Times investigation revealed that the president is making millions of dollars through his national security dealings.

The Daily on September 24, 2025 in New York Times

The United States agreed to give the UAE access to “hundreds of thousands of the crown jewels of American technology,” the most advanced AI chips. At the same time, $2 billion from the UAE went into World Liberty, the cryptocurrency company created by Donald Trump and Steve Witkoff. In effect, the UAE gets the chips while Trump’s family and allies get paid.

Read article

AI and Elections

Meta Ramps Up Spending on A.I. Politics With New Super PAC

Eli Tan and Theodore Schleifer on September 23, 2025 in New York Times

Meta is taking a more combative stance on AI policy, launching two super PACs—Meta California and the newly announced American Technology Excellence Project—to shape political outcomes in favor of AI industry growth. Backed by “tens of millions” in initial spending, the new PACs will target state-level politicians seen as barriers to tech advancement. Led by bipartisan strategists, the move signals a shift from Meta’s traditionally limited political engagement toward high-stakes electoral influence, especially ahead of the 2026 midterms. With over 1,100 AI-related bills introduced in U.S. states this year, the tech industry is mobilizing to define the regulatory landscape, putting AI politics on a collision course with state governance.

Read article

AI and Public Safety

Making Better Access to Law Enforcement

Editoral Board on September 15, 2025 in Marietta Times

Ohio has become the first state to deploy an AI-infused public safety app designed to streamline anonymous tips to law enforcement. The new Safeguard Ohio app allows users to upload photos, videos, and audio while AI prompts help guide the user to provide detailed information—automatically following up until the tip is complete. Officials hope the app will lower barriers to reporting and improve public safety intelligence. But the Marietta Times editorial warns of risks: false accusations, misuse in polarized environments, and the challenge of ensuring equity and discretion in AI-guided interactions. Still, the tool could improve access for residents hesitant to engage directly with law enforcement, if it’s deployed with accountability and care.

Read article

AI and Public Engagement

Reward, Risk, and Regulation: American Attitudes Toward Artificial Intelligence

Staff on September 17, 2025 in Gallup and the Special Competitive Studies Project

A first-of-its-kind national survey finds that while nearly all Americans have heard about AI, only 8% feel “very knowledgeable” and just 39% use it regularly. Conducted by Gallup and SCSP, the report shows strong bipartisan support for AI oversight, including independent testing and workforce training, but persistent skepticism around fairness, job impacts, and national competitiveness. Only 15% believe the U.S. is likely to lead globally in AI a decade from now. Support for international cooperation outweighs unilateral approaches, and trust increases when AI is framed as a defensive tool. The report offers a baseline for understanding U.S. public opinion on AI regulation, risk, and leadership.

Read article

AI and Problem Solving

Choosing Wisely: A New Resource for Picking the Right Participation Platform

Dane Gambrell on September 23, 2025 in Reboot Democracy Blog

Choosing a digital participation platform is a governance choice. Dane Gambrell's latest post explores what the new People Powered’s Guide to Digital Participation Platforms gets right, the idea that participation only works when tools are matched to purpose, capacity, and follow-through.

Read article

▲