News

Artificial intelligence is already reshaping democracy — and not in subtle ways. That is the central message of Rewiring Democracy, the new book by security technologist Bruce Schneier of Harvard Kennedy School and data scientist Nathan Sanders, and the focus of my recent conversation with Bruce on the Economics Explored podcast.

The book’s core claim is striking: AI’s impact on democracy will be far larger than that of social media. The reason is simple. Democracy is a deeply human institution — built on human judgement, persuasion, administration, and participation — and AI is increasingly embedded in all of those processes.

This is not a future problem

One of the most important takeaways from Rewiring Democracy is that AI is already affecting democratic systems, often invisibly. Of course, we are still in the early years of AI adoption, and so far, there has been a mix of successes and failures in AI experiments.

To illustrate, AI tools are now being used in:

• Political campaigns, to optimise messaging, polling, and “get out the vote” efforts (e.g., Cornell researchers tested AI chatbots programmed to advocate for specific candidates in controlled experiments conducted during the 2024 U.S. presidential election, the 2025 Canadian federal election, and the 2025 Polish presidential election, finding that brief interactions with these chatbots could shift voters’ opinions by measurable margins). (1)
• Legislatures, to help draft legislation, analyse how new laws interact with existing statutes, and assess regulatory impacts (e.g., in the UK, AI-enabled legal analysis tools are being piloted to identify conflicts, overlaps, and unintended consequences between proposed laws and the existing statute book before parliamentary debates). (2)
• Public administration, from audit targeting and compliance monitoring to routine decision support (e.g., in Amsterdam, city officials deployed an AI-driven welfare fraud detection system called “Smart Check” designed under responsible-AI guidelines, although when piloted on real welfare applications, it continued to produce biased outputs and was no more effective than human caseworkers, leading the municipality to terminate the project). (3)
• Courts, assisting with legal research, interpretation of language, and litigation strategy (e.g., at a 2025 Southern California Association of Law Libraries panel, law librarians compared AI-driven legal research platforms, such as Lexis+ AI, Westlaw Precision AI, and vLex’s Vincent AI, and found that while all could quickly answer basic legal questions, each showed distinct strengths and limitations and should be treated as a starting point rather than a definitive source for legal research). (4)
• Citizen oversight, where AI systems help journalists and civil society identify anomalies in political behaviour that warrant scrutiny (e.g., operation Serenata de Amor in Brazil illustrates how AI enables new forms of civic engagement by using machine learning to detect suspicious political expense claims and automatically publicising them via a Twitter bot to prompt public scrutiny and political accountability). (5)

Individually, many of these uses look incremental. Collectively, they amount to a fundamental change in how democratic systems function.

AI amplifies power — for better or worse

Schneier’s argument is not that AI is inherently pro- or anti-democratic. Instead, AI amplifies existing power structures.

If democratic institutions are strong, accountable, and well governed, AI can help them work better — faster administration, better-informed citizens, and more inclusive participation. If institutions are weak, captured, or drifting toward authoritarianism, AI can just as easily accelerate that drift.

This is why Schneier repeatedly emphasises that the biggest risks are not technical, but institutional and economic. Today’s most powerful AI systems are controlled by a small number of very large technology firms, operating under incentives that prioritise commercial returns rather than democratic outcomes.

From an economic perspective, this is familiar territory: scale economies, network effects, and market concentration combined with weak regulation rarely deliver socially optimal outcomes.

Rewiring democracy, not abandoning it

Where Rewiring Democracy is particularly strong is in rejecting fatalism. The book argues that AI also creates opportunities to rethink and strengthen democratic processes.

One example is citizen participation. AI can dramatically lower the cost of organising, moderating, and synthesising large-scale public deliberation. Experiments with AI-facilitated citizen assemblies — involving hundreds or thousands of participants in countries such as Germany and Scotland — would have been inconceivable using traditional methods.

This is not about replacing representative democracy, but about augmenting it: giving citizens more meaningful ways to engage, and giving decision-makers better insight into public preferences and trade-offs.

In economic terms, AI changes the scale economics of deliberation and participation.

The case for public AI and regulation

Schneier is clear that better outcomes will not emerge automatically. Two ideas stand out in his view.

First, the concept of public AI: AI systems developed transparently, without corporate control, and explicitly for public benefit. A recent Swiss government–university partnership, the Swiss government–ETH Zurich model, shows this is feasible without the astronomical costs often claimed by industry.

Second, the need for regulation. Transparency alone is not enough; it must be paired with accountability. AI systems that can influence behaviour, shape information environments, or make consequential decisions should be treated like other powerful technologies — regulated to manage risk, limit abuse, and curb excessive concentration of power.

Antitrust policy, in particular, matters. Democracies function poorly when economic power becomes too concentrated, and AI raises the stakes.

Why this matters

The key message for policymakers, economists, and citizens is that AI is already rewiring democracy. The question is whether we shape that process deliberately or allow it to be driven by narrow commercial incentives.

Rewiring Democracy is not a book about technology alone. It is about power, institutions, and governance — and it deserves the attention of anyone concerned with the future of democratic societies.

You can listen to my full conversation with Bruce Schneier on the Economics Explored podcast, where we explore these ideas in more depth.