Google AI Tested on 4,480 Election Queries in Germany

AlgorithmWatch, a Berlin-based nonprofit that tracks algorithmic systems, ran 4,480 search queries through Google to see how the company's AI Overviews handle election-related questions. The test window covered three 2026 votes in eastern Germany: state elections in Saxony-Anhalt and Mecklenburg-Vorpommern, plus the Berlin House of Representatives race. Access came through Article 40(12) of the EU's Digital Services Act, which lets vetted researchers query platforms directly via Google's Search Researcher Result API rather than scraping results by hand.

The group wanted to know three things: how often an AI Overview appears for political searches, where the underlying sources come from, and how the summaries describe candidates and parties. Earlier reporting on AI search tools has flagged bias and sourcing problems in general terms, but AlgorithmWatch's dataset lets it quantify the pattern for a specific set of German races, tracking answers for the far-right AfD against other parties and counting how concentrated the linked domains are. The findings raise questions about how much control Google exercises over when these summaries surface at all, and what shows up when they do.