Beth Simone Noveck

Who Will Shape AI in the Public Interest The current controversy over the Pentagon’s AI contracts reveals a deeper issue: governments are shaping the AI market through procurement in the wrong ways or not at all, failing to make demands that...

Recent tensions between the U.S. Dept of Defense and Anthropic highlight that the government isn’t just regulating AI, it’s one of its largest customers.

Procurement shapes markets. Will we shape AI by default or by design?

Read more: rebootdemocracy.ai/blog/public-ai

Research Radar: Academics Are Sounding the Alarm on AI Adoption. Who's Listening? A new paper urges universities to resist the uncritical adoption of AI, arguing that existing research integrity standards already prohibit much of what institutions are normalizing. Beth Simone Novec...

Universities are rapidly adopting commercial AI tools.

A new paper warns that research integrity standards already raise concerns about transparency and independence. For many in the public sector, refusal isn’t realistic. Where do we go from here?

Read more: rebootdemocracy.ai/blog/researc...

AI for Governance: How Institutions Can Provide AI Access Safely, Affordably, and at Scale As governments move from AI pilots to workforce-wide access, the hardest questions are practical: What does it cost at scale? Who gets access? How do you log usage and manage risk? How do we pair acce...

Enterprise AI access doesn’t have to mean multi-million-dollar-per-seat contracts. NJ built inside its own cloud (~$1/user/month). Boston’s pilot is under $10K year one.

Pricing structure matters as much as model performance.

Read today's #AIforGovernance piece: rebootdemocracy.ai/blog/how-bos...

Reboot Weekly: Building a Culture of Recognition, AI Safety by Local Rule, and Argentina’s Strategy Reset - News That Caught Our Eye This week on Reboot Democracy, Max Stier explores how AI can strengthen internal government culture by building real-time recognition systems for civil servants. Elana Banin examines the UbuntuGuard b

📖 Read this week’s #NewsThatCaughtOurEye: rebootdemocracy.ai/newsthatcaug...

Evaluating AI Safety Through Local Policy: Findings from the UbuntuGuard benchmark A new paper introducing the UbuntuGuard benchmark questions whether strong results on English-language safety tests consistently translate into responsible use. Built from policies developed by 155 Af...

AI safety isn’t universal.

#UbuntuGuard, led by researchers at @brownpublichealth.bsky.social and collaborators, shows that models that pass English benchmarks can fail when tested against locally grounded policy rules across African languages.

Read more:
rebootdemocracy.ai/blog/ubuntu-...

Minding the brand: Leveraging AI to build a culture of recognition in government Max Stier argues that rebuilding trust in government begins inside agencies by recognizing and elevating the everyday excellence of career civil servants. Drawing on the Partnership for Public Service...

Presidents’ Day honors those who have held the nation’s highest office.

But trust in government is built inside agencies day to day.

Max Stier of @ourpublicservice.bsky.social explores how AI can strengthen cultures of recognition and performance.

Read more: rebootdemocracy.ai/blog/minding...

Global AI Watch: Korean Public Funds for Global AI Advancements As Korea aims to become a top-three global AI power, the government has staked its strategy on “sovereign AI” as both an economic and national security priority. In this reflection, originally publish...

In our latest #GlobalAIWatch, @hickokmerve.bsky.social lexamines South Korea’s $73.5B public investment in a national foundation model and asks whether smaller language models, energy efficiency, and stronger evaluation may be wiser bets.

Read: rebootdemocracy.ai/blog/korea-p...

How We Co-Designed an AI-Powered Tool for IEPs with Families in San Francisco As the AIEP project concludes its first pilot in San Francisco, it offers more than a new AI tool for navigating IEPs. It shows what becomes possible when families, educators, designers, and researche...

150+ parents helped co-design an AI tool to make IEPs clearer and more accessible.

As #AIEP’s San Francisco pilot wraps, we’re sharing what it reveals about building AI with communities.

📖 Read the blog:
rebootdemocracy.ai/blog/co-desi...

🗓️ Workshop @ 2:00 PM ET:
innovate-us.org/human-center...

Doing Democracy with AI: Designing Public Engagement for the AI Era Leaders increasingly believe public engagement matters, but lack the practical know-how to do it well. Beth Simone Noveck and Dane Gambrell examine how institutions use AI and collective intelligence ...

Public engagement matters. Designing it well is hard.

Dane Gambrell and I co-wrote a new #AIforGovernance blog on how AI can support participation.

We’re building a course, "Designing Democratic Engagement for the AI Era," and seeking feedback.

rebootdemocracy.ai/blog/doing-d...

Wicked Decluttering What if the problem with government wasn’t too many rules, but how they’re organized? Boston’s permitting overhaul with AI for Impact shows how AI and collective intelligence can simplify the user exp...

From insights at the @opengovpartnership.org Summit to @boston.gov, @mayorwu.boston.gov redesigned permitting to show how wicked decluttering can use AI to reorganize rules around human intent—without weakening safeguards.

🔗 rebootdemocracy.ai/blog/wicked-...

Research Radar: “Unboxing the Prompt”: How Community Feedback (and AI) Helped Us Build Better AI Together Families are expected to advocate for their children using IEP documents that are dense, technical, and often inaccessible. Instead of treating AI as a black box that produces generic summaries, this ...

Early AIEP prototypes “worked” but failed where it mattered, developing polished summaries that didn’t answer the real questions parents need to advocate for their kids.

That failure reshaped the tool by putting families in the loop.

rebootdemocracy.ai/blog/unboxin...

Experimentation as Public Infrastructure Governments are adopting powerful new technologies faster than their systems are built to learn. This piece by Cassandra Madison at the Center for Civic Futures argues that responsible innovation requ...

Governments don’t fail because they experiment. They fail because they don’t.

Cassandra Madison argues governments need safe places to fail early, so AI systems don’t fail at scale.

Experimentation isn’t a side project. It’s public infrastructure.

📖 rebootdemocracy.ai/blog/experim...

Mapping the School, Seeing the System: How Spatial Context Reshaped Public Decision-Making in Uzbekistan and Bhutan Mapping schools in Uzbekistan and Bhutan revealed how data that appeared complete on paper often missed what mattered on the ground, like distance, access, and everyday conditions. By treating geograp...

#GlobalAIWatch: Mapping schools in Uzbekistan and Bhutan showed how data that looks complete on paper can miss what matters on the ground.

Spatial data, participation, and AI helped connect evidence to real decisions. Not louder data, just clearer.

🔗 rebootdemocracy.ai/blog/mapping...

Prediction Isn’t Intelligence: How Predictive Models Really Work in Government A hiring tool changes your "personality" score based on whether there's a bookshelf behind you in an interview. A hospital model suggests asthmatics are safer—because it learned how to respond to the ...

At the airport, you look into a camera, get a green checkmark, and walk through.

That moment depicts both the promise and risk of AI in government. New #RebootDemocracy post featuring @randomwalker.bsky.social on why predictive models need governance.

📖 rebootdemocracy.ai/blog/predict...

Launching the Observatory of Public Sector AI: An Invitation to Build the Evidence Base Together We’re launching the Observatory of Public Sector AI, a new research initiative of InnovateUS and The GovLab that draws on data from more than 150,000 public servants nationwide. By analyzing how publi...

We launched the #ObservatoryofPublicSectorAI, one of the largest datasets to date on how public servants learn, use, and adapt AI at work.

Our goal is to build the evidence on what strengthens government capacity.

Learn more: rebootdemocracy.ai/blog/launchi...

Reboot Weekly: Modernizing Regulation, Making Services Multilingual, and Listening First - News That Caught Our Eye Reeve Bull shows how Virginia is using AI to improve regulatory review, from analyzing statutes to streamlining outdated rules. Anirudh Dinesh looks to India, where the Bhashini platform expands democ

📖 Read this week’s #NewsThatCaughtOurEye: rebootdemocracy.ai/newsthatcaug...

Voices in Every Language: How India is Building More Inclusive AI India's Bhashini platform is democratizing access to digital services for 1.4 billion people by treating multilingual capability as public infrastructure. Through crowdsourced voice donations and open...

Language is one of the most underestimated barriers to democratic access.

The Government of India's #Bhashini program shows what’s possible when multilingual AI is treated as public infrastructure, so people can be understood by the systems meant to serve them.

rebootdemocracy.ai/blog/voices-...

AI and the Future of State Regulation This piece examines how the Commonwealth of Virginia is using artificial intelligence to modernize regulatory review, shifting the focus from regulating AI to governing with it. Drawing on recent refo...

The #CommonwealthofVirginia is using AI to modernize regulatory review, including analyzing statutes, reducing duplication, and clarifying rules.

On the #RebootDemocracy blog, the “Virginia Model” shows how other states can strengthen regulatory capacity with AI.

rebootdemocracy.ai/blog/ai-and-...

The NLRB Has a Data Problem. One Lawyer Is Using AI to Fix It. U.S. labor law is shaped by thousands of decisions issued each year by the National Labor Relations Board (NLRB) and the courts, yet much of this precedent remains difficult for workers, organizers, a...

#ResearchRadar: Access to the law shapes whether rights exist in practice.

This piece examines how fragmented NLRB data limits workers' power and how NLRB Research uses AI and open-source tools to make labor law more usable.

Read more: rebootdemocracy.ai/blog/nlrb-da...

Using AI to Improve Public Services in New Jersey: An interview with Dave Cole New Jersey’s Office of Innovation has received a Public Benefit Innovation Fund grant to expand its AI platform with tools that help residents access benefits faster and with fewer errors. In this con...

How does AI move from pilot to public infrastructure? In a new #RebootDemocracy interview, Dave Cole unpacks how New Jersey is scaling responsible AI across agencies backed by the Public Benefit Innovation Fund.

🔗 rebootdemocracy.ai/blog/ai-publ...

Can mid-sized economies come together to build frontier AI? Conventional wisdom presents mid-sized economies with two options for accessing advanced AI: rely on American or Chinese systems, or fall behind. Neither choice preserves the technological sovereignty...

Interesting read from the @oecd-ocde.bsky.social l: Charles Martinet & Yohann Ralle explore whether shared compute and cooperation could help mid-sized economies build frontier AI while retaining sovereignty.

Worth a look: rebootdemocracy.ai/blog/mid-siz...

Research Radar: The City as Mesh and New Ways of Organizing for Effective Problem Solving The City as Mesh by Geoff Mulgan and Caio Werneck offers a powerful new framework for organizing cities to tackle cross-cutting challenges. This Research Radar examines their new paper and argues for ...

What’s the right way to organize government for real problem-solving?

A background paper by Geoff Mulgan and Caio Werneck argues that cities need flexible ways of working across boundaries.

Join the Bloomberg Center for Cities tomorrow for the conversation.

rebootdemocracy.ai/blog/researc...

From Red Tape to Green Tape: Decluttering the State with AI and Collective Intelligence Governments are increasingly using AI to identify redundant, outdated, and burdensome regulations. But efficiency alone is not reform: without public judgment, simplification can weaken essential prot...

Governments from Virginia to Ohio to San Francisco are using AI to cut regulatory clutter.

But efficiency alone isn’t reform.

In From Red Tape to Green Tape, I argue AI must be paired with collective intelligence.

🔗 rebootdemocracy.ai/blog/green-t...

Research Radar: How Research Drove a Community-Led Action Plan for LA Wildfire Recovery This Research Radar examines how California’s Office of Data and Innovation used a rigorous, community-driven approach to translate thousands of survivor contributions from the LA wildfires into a con...

How can research turn community input into real policy?
This #ResearchRadar shows how CA used deliberation, digital tools & AI to turn wildfire survivor input into a recovery action plan.

📖 rebootdemocracy.ai/blog/researc...

🎥 innovate-us.org/building-a-b...

Using AI to Support Public Deliberation: A Conversation with Audrey Tang In this workshop, Audrey Tang and Danielle Allen discuss how AI-enabled civic technologies, paired with radical transparency and thoughtful institutional design, can help democracies respond to proble...

After months of learning in our InnovateUS series, the lesson is democracy works when governments can listen, learn, and act on public input.

From Taiwan to California, it shows engagement tied to decisions reduces polarization and builds trust.

rebootdemocracy.ai/blog/deliber...

Reboot Weekly: Testimony, Training, and the Infrastructure of Democracy - News That Caught Our Eye Beth Simone Noveck testifies before Congress on how AI can help lawmakers move beyond performative public engagement to use constituent input at scale meaningfully. Agueda Quiroga announces the launch

📖 Read our weekly #NewsthatCaughtOurEye:

rebootdemocracy.ai/newsthatcaug...

YouTube video by Committee on House Administration Subcommittee on Modernization and Innovation: “The Future of Constituent Engagement with Congress”

Today I testified in front of the House Subcommittee on Modernization and Innovation on the Future of Constituent Engagement

📍 Live testimony www.youtube.com/watch?v=OhYk...

📌 Topic: The Future of Constituent Engagement with Congress rebootdemocracy.ai/blog/future-...

Public AI: Policies for democratic and sustainable AI infrastructures This analysis by OECD.AI contributors offers a clear, practical blueprint for “public AI.” It maps where power concentrates across the AI tech stack, such as compute, data, and models, and shows how g...

@oecddigital.bsky.social contributors outline a practical blueprint for public AI, mapping power across compute, data & models, and showing how policy can reduce dependence on frontier labs through a gradient of democratic, public-interest infrastructure.

📖 rebootdemocracy.ai/blog/public-...

Learning Together to Improve Public Service Since its creation in 2023, InnovateUS’s training offerings have supported public professionals in making sense of change together, building capacity, and strengthening their ability to make informed ...

🔔 #InnovateUS Spring 2026 workshops are live

Next semester is shaped by listening to what helps in practice: shared learning, practical judgment, and skills grounded in public service.

Reflection by Agueda Quiroga: rebootdemocracy.ai/blog/learnin...

Research Radar: Synthetic Data Is Redefining Representation As governments experiment with AI to simulate public opinion, new questions are emerging about who these systems truly represent. This Research Radar examines the Collective Intelligence Project’s Dig...

New #ResearchRadar: a look at @cip.org Digital Twin Evaluation Framework and what it reveals about the limits of letting AI speak for people.

AI models are starting to act as stand-ins for real communities. But can synthetic publics actually represent us?

rebootdemocracy.ai/blog/researc...