Inside the Campaign on X: AfD Surge, Pro-Russian Framing, and Saxony-Anhalt 2026
UncovAI evaluated 7,612 posts on X ahead of the September 6, 2026 state election in Saxony-Anhalt. Our findings reveal heavily skewed endorsement patterns, strong foreign-policy weaponization, and specific disinformation vectors designed to undermine public trust prior to voting day.
Executive Summary & Key Takeaways
- Asymmetric Mobilization: Of 215 posts expressing an explicit voting endorsement, 149 favored the AfD (69.3%) compared to 66 for all competitors combined (30.7%).
- Nationalized State Campaign: Foreign policy and federal issues dominated the regional race. Topics like war/peace (5.2%), Russia (4.8%), and migration (4.1%) heavily outranked local governance.
- Core Disinformation Vector: The Leipzig/Halle drone incident was reframed across accounts as a "false flag" to prevent an AfD government—a claim that collapsed after German security authorities formally attributed the attack to Russia on September 1–2, 2026.
- Pre-Bunked Fabrications: Viral claims of a "44% Prognos poll" were verified false (Prognos AG does not run political polling). Claims alleging mass postal-vote manipulation were circulated entirely without evidence.
- Organic Multiplier Network: The dataset reflects high-velocity political amplification and aligned messaging rather than a centralized, technical bot army.
1. Endorsement Share: AfD vs. Other Parties
Filtering strictly for posts containing a clear call to vote or explicit candidate endorsement (n = 215) isolates active partisan mobilization from general commentary:
| Alignment | Posts | Share of Endorsements |
|---|---|---|
| Pro-AfD | 149 | 69.3% |
| Pro-CDU | 49 | 22.8% |
| Pro-Linke | 7 | 3.3% |
| Pro-BSW | 4 | 1.9% |
| Pro-SPD | 3 | 1.4% |
| Pro-Greens | 2 | 0.9% |
| Pro-FDP | 1 | 0.5% |
| All Competitors Combined | 66 | 30.7% |
Pro-AfD endorsements outpaced all other political parties combined by 2.26 to 1 within explicit recommendation posts.
2. Primary Narrative Clusters
"Historic AfD Breakthrough"
Positions Saxony-Anhalt as the staging ground for the first AfD state premier (Ulrich Siegmund), leaning on landslide rhetoric.
"AfD = Peace" Framing
Casts Berlin, NATO, and the EU as belligerents while packaging a vote for the AfD as an anti-escalation corrective.
Anti-Coalition Warnings
A tactical voting message aimed at CDU supporters, claiming a vote for the CDU inevitably yields a left-wing coalition.
Institutional & Media Bias
A closed loop framing media criticism of the AfD as proof of establishment fear and campaign legitimacy.
3. Fact-Check Summary
| Claim Tracked | Finding | Status |
|---|---|---|
| Leipzig/Halle drone attribution was a "false flag" | Attributed officially to Russia by federal investigators on Sept 1–2, 2026. Staged-claim lacks evidence. | Debunked |
| "AfD polling at 44% per Prognos" | Prognos AG does not conduct voter polls and disavowed the originating account. | False |
| Postal voting is rigged to alter results | Preemptive delegitimization effort circulated without empirical proof or documentation. | Unsupported |
| AfD guaranteed an absolute majority | Strong polling (42–43%) does not mathematically guarantee a single-party seat majority. | Unverified |
Frequently Asked Questions
What is the scope of this dataset?
7,612 posts on X gathered and classified up to September 2, 2026, evaluating reach metrics, narrative framing, and party endorsements ahead of the September 6 vote.
Does this data prove a Russian-directed botnet?
No. While narrative themes heavily overlap with known pro-Russian information environments, amplification patterns point primarily to active political multiplier accounts and organic retweets rather than technical bot coordination.
Is the raw dataset available for public download?
No. UncovAI does not make the raw structured dataset publicly downloadable. Research institutions, newsrooms, and verification desks can request access through direct inquiry.
Need Access to This Dataset?
UncovAI provides structured intelligence, reach data, and custom text/content provenance analysis for newsrooms, academic bodies, and institutional investigators. We do not distribute raw data files publicly.
If you need this dataset — write to us →