AI also lost the election

Summary of AI also lost the election

by Vox

26m•November 10, 2026
Listen to Episode

Overview of AI also lost the election

This Vox Today, Explained episode argues that AI—especially the rapid buildout of data centers and the broader policy vacuum around frontier models—took a political hit in the recent midterms. The show frames the election as a warning sign for Big Tech: voters rejected many AI/data-center-related measures, Democrats won in places where they opposed these projects, and both parties are now under pressure to move from vague promises to actual regulation. The episode features two key voices: New York Assemblyman and AI safety advocate Alex Bores and Chris Lehane, OpenAI’s chief global affairs officer.

What the episode is about

  • The episode starts by noting that AI-related ballot measures and data-center initiatives were on the ballot more than 40 times across 10 states, and all but one were defeated.
  • It argues that the public is increasingly skeptical of AI infrastructure and of the companies behind it.
  • The central question: What should Democrats and policymakers do now that AI is becoming a major political issue?

Key takeaways

Voters want action, not just rhetoric

Alex Bores argues that voters are no longer impressed by broad statements about AI safety. His view:

  • Neither party is trusted on AI.
  • Democrats should use their new political power to pass concrete rules.
  • The public wants solutions to multiple AI-related harms at once, not a forced choice between catastrophic-risk concerns and near-term issues like privacy, labor, deepfakes, and child safety.

AI regulation does not necessarily kill innovation

Bores pushes back on the idea that strong AI rules would crush the economy:

  • He says most sensible regulation would function more like “a seatbelt in a Lamborghini” than a shutdown.
  • He argues that the industry exaggerates compliance burdens.
  • In his view, smart regulation can protect people without stopping AI development.

OpenAI wants mandatory safety rules too

Chris Lehane presents OpenAI as supportive of national-level AI safeguards:

  • He says OpenAI has long supported mandatory safety requirements for frontier AI labs.
  • He frames the election as a sign that the politics are now aligning with the policy.
  • He also argues that regulation could help create international standards.

Policy ideas discussed

Alex Bores’ main proposals

Bores lays out three broad buckets of policy action:

  1. Common-sense guardrails

    • Require third-party auditors at frontier AI companies with employee-level access.
    • Require disclosure of safety incidents.

  2. Technology investments

    • Fund technologies that can help verify or limit what chips and models are doing.
    • He compares this to verification systems used in other high-stakes domains, like nuclear oversight.

  3. Buy time for safety research

    • Slow things down enough to determine whether advanced AI can be made safe.
    • He singles out full recursive self-improvement—AIs building the next generation of AIs without humans in the loop—as the one thing he most wants banned immediately.

Lehane’s response

Lehane says OpenAI would support something like:

  • Broadly defined speed limits on AI development.
  • A framework that ensures humans remain in control.
  • Not an absolute yes/no on recursive self-improvement, but a system that keeps AI from accelerating beyond human oversight.

Data centers, “compute campuses,” and public backlash

A major portion of the episode focuses on OpenAI’s effort to reframe data centers as “compute campuses.”

Lehane’s argument

He says the term is intentional because:

  • These facilities are not just storage sites; they are where compute happens.
  • They support both model training and inference.
  • They can create real economic benefits, including high-paying technical jobs.

The public-relations problem

The episode highlights the backlash:

  • Many communities oppose data centers because of concerns about land use, power consumption, local control, and broader distrust of Big Tech.
  • Lehane acknowledges the distrust but argues that companies need to listen, learn, and build with communities, not impose projects on them.
  • He says state and local rules should ensure communities see a real material benefit.

Why 2028 may be the real AI election

Both guests suggest AI will be much bigger politically in the next presidential cycle.

  • Bores says 2026 is a trailer for 2028.
  • Lehane agrees that AI could become the issue in 2028.
  • The episode suggests both parties will need a real answer to a larger question:
    • Who controls the future economy, and how do ordinary people maintain agency?

The bigger political theme is that candidates can’t just run as “against things.” They’ll need:

  • A positive vision for the social contract,
  • A plan for how AI fits into the American dream,
  • And a credible answer to the public’s anxiety about losing control.

Bottom line

The episode’s core message is that AI is no longer just a tech story—it’s now an election issue. Voters are showing skepticism toward AI infrastructure and unchecked development, while policymakers are starting to push for:

  • mandatory safety standards,
  • community protections,
  • stronger oversight of frontier labs,
  • and a slower, more controlled path for advanced AI.

The takeaway from both interviews is similar: AI is moving fast, but political legitimacy will depend on whether humans stay in control.