< formatted article >

The Hidden War Over Public Comments: How AI is Fueling Astroturfing in Policy Debates

Government agencies have long relied on public comments to gauge voter sentiment—whether it’s climate regulations, energy projects, or local ordinances. But a troubling trend has emerged: sudden, coordinated floods of identical emails, all pushing the same agenda—often in favor of fossil fuels or against green restrictions.

Take the South Coast Air Quality Management District in California, where an unexpected email surge arrived just before a vote on banning gas heaters. Most messages bore real names, but some signers denied ever sending them.

The Middlemen Behind the Messages

Behind these campaigns often lurk digital intermediaries—companies like CiviClick, which specialize in mass email blasts for clients. They promise "personalized" letters, sometimes using AI to generate them at scale. Yet many insist human hands still hit "send."

But verification raises questions. When officials checked 172 comments, only five responded. Two admitted sending them; three said no. The silence speaks volumes.

AI’s Role in Astroturfing: When Corporate Campaigns Pose as Grassroots Movements

Critics argue that AI-powered mass messaging is corrupting public debate. It’s not that AI itself is malicious—it’s how effortlessly it enables astroturfing, where corporate-backed campaigns masquerade as genuine public outcry.

In New York, union-backed emails flooded in supporting natural gas, some arriving hours late. In North Carolina, energy giant Williams Cos. used similar tactics to push for pipeline approval—until officials spotted odd patterns, like messages sent at 3 AM from similar email addresses.

Some agencies are fighting back. California’s Bay Area Air District is rolling out a new portal requiring real-name verification before accepting comments. Meanwhile, in Berkeley, city officials were blindsided by an AI-driven campaign that generated seemingly personal emails in favor of surveillance cameras.

Users clicked a link, answered a few questions, and—instantly—a draft email appeared in their inbox, ready to send. The wording varied, but the pro-surveillance message was unmistakable.

The Human Factor: Can AI Write the Email, But People Still Cast the Vote?

Despite AI’s growing influence, human action remains decisive. In Berkeley, the council rejected the camera plan after in-person opposition from residents. AI may craft the message—but people still hold the power.

As digital manipulation grows more sophisticated, one question looms: How can public input stay genuine in an era of AI-driven astroturfing?