One account, most of the attention
A week before Saxony-Anhalt's September 6, 2026 state election, one politician's TikTok account absorbed nearly two-thirds of all political attention on the platform in the state. The University of Potsdam's Social Media Monitor tracked 96 political TikTok accounts in Saxony-Anhalt that posted 2,227 videos between June 1 and August 31, 2026, drawing 43.4 million views combined. AfD lead candidate Ulrich Siegmund's account alone drew 28.0 million of them — 64.5% — averaging roughly 214,000 views per post against Minister-President Sven Schulze's (CDU) roughly 5,600 (Universität Potsdam). By comparison, the leading accounts in the two other states that voted in September 2026, Mecklenburg-Vorpommern and Berlin, took just 13% and 9% of their respective political TikTok attention — Saxony-Anhalt's concentration was an outlier (netzpolitik.org). On election day, the AfD won 43.8% of the vote and 39 of 83 seats — first place, three short of a governing majority (Al Jazeera).
The case for scrutiny
The instinct to ask whether TikTok's recommender system is doing this to Germany, rather than for it, is not paranoid. The precedent is real: in December 2024, Romania's constitutional court annulled the first round of its presidential election after intelligence findings that coordinated activity on TikTok had manipulated visibility for a fringe candidate, triggering a European Commission investigation into whether TikTok's recommender system was gamed (EDRi). Germany's regulator has taken that risk seriously since: the Bundesnetzagentur, as the country's Digital Services Coordinator under the EU's Digital Services Act (DSA), ran a formal stress test with TikTok and six other major platforms in January 2025 specifically to rehearse detection of election-related systemic risks — illegal content, foreign interference, and manipulation of information (Bundesnetzagentur). A separate Bertelsmann Stiftung experiment on the 2025 federal election found that TikTok surfaced #afd-tagged videos to fresh accounts within 11-12 minutes of posting, versus about 70 minutes for SPD-tagged content — a gap researchers said could reflect the platform treating AfD material as unusually "reactive" content worth pushing harder (Bertelsmann Stiftung). If a recommender system is structurally amplifying one party's message over others', that is exactly the kind of systemic risk the DSA's Article 34-35 risk-assessment regime and Article 38 non-profiling-feed requirement were built to catch.
Why the Saxony-Anhalt numbers don't prove that
But the Potsdam researchers are unusually disciplined about their own limits, and Germany's post-NetzDG regulators should be too. The study's own methodology note says the view counts "say nothing about causes — neither about the recommendation algorithm nor about content or paid reach" (Potsdam Social Media Monitor). Researcher Johannes Wolfgram's explanation for the concentration was mundane: existing audience size, not posting frequency or platform favoritism, predicts reach — the CDU published nearly three times as many posts as Siegmund and still captured a fraction of the views. Siegmund built an audience of over 820,000 TikTok followers over several years of native short-form content; Schulze did not. A large following that a recommender system then serves efficiently to existing subscribers is a different regulatory object than a recommender system independently boosting one party's fringe content over another's, which is what the Romanian case turned on. Conflating the two would repeat the core flaw of Germany's old Netzwerkdurchsetzungsgesetz (NetzDG), which the DSA formally superseded in February 2024 precisely because take-down-driven, platform-specific rules proved clumsy next to systemic-risk oversight (HateAid) — NetzDG punished platforms for content outcomes without ever establishing why those outcomes occurred.
The actual regulatory lever is boring, and that's the point
The honest answer is that nobody, including the researchers who produced these numbers, currently knows whether Siegmund's dominance reflects algorithmic favoritism, superior organic content strategy, or simply years of head start building a following before rivals took the platform seriously. That uncertainty is itself the strongest argument for the DSA's Article 40 researcher-data-access regime, which opened its EU-wide application portal in October 2025 specifically so vetted researchers could examine recommender-level data that public view counts cannot reveal — and which the Commission has already opened proceedings against TikTok over, for compliance shortcomings. If Berlin's political class wants to know whether TikTok is thumbing the scale for the AfD, the fix is pushing for full compliance with data-access obligations already on the books, not reaching for content restrictions on unproven causal claims. A regulator that moves against a party's speech based on a popularity metric its own source study disclaims as inconclusive would hand the AfD a legitimate grievance, and hand every future Bundestag majority a precedent for silencing whoever is next to build an audience the incumbents didn't see coming.