Google to muzzle its AI bots ahead of the 2024 election

Big tech has spent the last decade taking the blame for election chaos, and Google is not waiting around to see how generative AI complicates things next year.

With multiple high-profile elections approaching—led by the U.S. presidential race—Google just announced a plan to restrict election-related queries across its generative AI platforms.

Pulling the plug on AI answers

The restrictions will apply directly to Google Bard and Search Generative Experience, the AI answers embedded right inside Google Search.

The changes are scheduled to roll out early next year. Google has not specified the exact prompt categories that will be blacklisted, but the company noted it plans to operate with an abundance of caution.

Instead of trusting generative models to accurately navigate policy stances or voting booth logistics, Google is essentially hitting the mute button on AI responses for sensitive civic topics.

Watermarks and campaign defenses

The AI restrictions are part of a broader transparency push across Google properties.

YouTube creators will soon be required to disclose when their videos feature altered or synthetic content. Google did not share how strictly this policy will be enforced, but it builds on rules already applied to political ads running on its networks.

Meanwhile, Google will keep routing election search traffic on Search, News, Maps, and YouTube to verified quality sources, while offering security defenses to campaigns through its Advanced Protection Program.

Competitors are handling this slightly differently. OpenAI recently rolled out better detection tools to spot disinformation inside its models, but it has stopped short of banning entire categories of election-related queries.

Why it matters

If you plan to use Bard or AI search early next year to compare candidates or look up voting information, expect to hit a wall. For users, that means generative search won't be your civic co-pilot just yet, forcing you back to classic blue links and established news outlets.

We think this is a smart, defensive move from Google. With hallucination issues still unsolved across the industry, being wrong about an election carries far more regulatory risk than being wrong about a movie review.

The challenge will be drawing the line. Restrict too aggressively, and users can't find benign polling information; restrict too loosely, and the model inevitably slips up. Furthermore, muzzling top-tier tools like Bard won't stop bad actors from using open-source models to flood platforms with synthetic political media.