On August 14, 2026, the European Values Center for Security Policy, in cooperation with the Czech-Taiwanese Business Chamber, organized the webinar “The Evidence Problem: What Democracies Can Prove About Election Interference — and What They Can Do About It.” Bringing together Rasťo Kužel, Mădălin Necșuțu, Dr. Eve Chiu, and Dr. Yu-Hsi Liu, the discussion examined how democracies can identify, attribute, and respond to foreign electoral interference when evidence is incomplete, time is limited, and overreaction may itself weaken public trust.
The event was held under the Chatham House Rule; the write-up below has therefore been anonymized and does not attribute individual remarks.
The Window Before Certainty
Election interference creates its greatest danger not when democracies fail to detect it, but when they cannot decide what to do with the evidence they have. Intelligence may be persuasive but undisclosable. Journalistic findings may be credible but legally insufficient. By the time attribution becomes courtroom-ready, the election may be over.
The webinar’s central argument was therefore not that democracies need perfect information. They need institutions capable of acting responsibly under uncertainty. That requires different evidentiary thresholds for public warnings, platform action, policy intervention, and prosecution—supported by trust, coordination, transparency, and democratic restraint.
Europe: When Institutions Fall Behind
The European cases showed what happens when interference outpaces institutions.
In Slovakia, a fabricated audio recording appeared two days before the 2023 parliamentary election, during the campaign moratorium. Its significance lay not only in its content but also in its timing: the recording attacked trust in the opposition, journalism, the Roma minority, and the election itself when meaningful political rebuttal was hardest to mount. Waiting for definitive attribution would have meant surrendering the information space.
Moldova, during the September 2025 elections, demonstrated a stronger model. Investigative journalists infiltrated Russian vote-buying networks, giving the public evidence that carried more credibility than official warnings alone. Authorities combined prosecutions, cyber defense, public education, and international assistance without resorting to measures that would have validated Russian claims of illegitimacy.
Romania offered another warning in 2025. Călin Georgescu’s sudden rise exposed the costs of weak anticipation. Annulment of the presidential election became the final safeguard—but also proof that once interference destabilizes trust, every available response causes democratic damage.
Taiwan: Influence Starts Offline
Taiwan’s experience shows why election interference cannot be reduced to removing false posts after they go viral. Chinese influence operations are embedded in relationships, institutions, travel, religious networks, and private communication channels long before they surface online.
The legal architecture rests in part on two statutes: the Civil Servants Election and Recall Act and the Presidential and Vice Presidential Election and Recall Act. Both prohibit the deliberate dissemination of false information intended to influence whether a candidate is elected. The 2024 presidential election became an important test of whether these provisions could move beyond paper safeguards and be applied to concrete cases involving fabricated polling, covert funding, and political mobilization linked to China. The experience also pushed authorities to strengthen reporting and coordination mechanisms ahead of future elections.
But law reaches only part of the problem. During the 2024 campaign, an AI-generated video alleged that William Lai had an illegitimate son. It originated from a Vietnamese account but circulated in Chinese—an example of how synthetic content, foreign infrastructure, and domestic political narratives can be combined to obscure responsibility and complicate attribution.
Research on Mazu temple networks exposes the deeper mechanism. Influence begins offline, through pilgrimages, subsidized exchanges, and interpersonal trust. It then migrates into private LINE groups, reshapes information sources, and eventually affects political behavior.
Academic findings may not be enough to convict anyone. They can still justify proportionate policy: closer scrutiny of foreign-funded exchanges, stronger transparency requirements, better coordination with platforms, and earlier engagement with local communities. Taiwan’s challenge is not simply to detect interference, but to disrupt it without allowing the response itself to erode democratic freedom.
One Playbook, Two Theaters
Russia and China do not need to run identical operations—or even coordinate every campaign—to reinforce one another.
In Moldova, Russia remains the immediate threat. It buys votes, cultivates political and religious proxies, launches cyberattacks, and seeks to delegitimize state institutions. China operates more quietly through telecommunications infrastructure, persistent cyber access, elite engagement, sponsored travel, and relationships with pro-Russian political forces. Russia plays first violin; China helps shape the hall in which the performance takes place.
In Taiwan, the hierarchy is reversed. China supplies the dominant political objective, local networks, narratives, and pressure. Russian methods and information frames can still serve as supporting instruments, particularly through amplification, anti-Western narratives, and shared techniques of institutional distrust.
The linkage is therefore less an alliance of command than an ecosystem of learning. Authoritarian actors borrow tools, narratives, and vulnerabilities across borders. Democracies, by contrast, too often treat each election as a national exception. That asymmetry gives the interferers an advantage.
Act Early—or Pay Later
The first lesson is that democracies should stop treating all evidence as though it must meet the standards of a criminal court. A suspicious amplification network may justify a public warning. Documented coordination may justify action by social media platforms. Research showing a durable influence mechanism may justify regulatory scrutiny. Prosecution rightly requires more.
The second lesson is that preparation must begin before the scandal. Yet, as noted during the discussion, prebunking carries an obvious limitation: Russia and China continually adapt their methods, while democracies risk preparing only for the last operation. The answer is not to predict every deepfake, platform, or proxy in advance. It is to monitor the narratives, communities, and institutional vulnerabilities that hostile actors repeatedly exploit. Longer-term monitoring can reveal attacks on postal voting, electoral authorities, alliances, religious identity, or social cohesion months before these narratives are converted into an election-day operation.
Prebunking should therefore focus less on guessing the precise form of the next manipulation and more on building public recognition of its underlying logic. Voters should understand how elections work, why sensational allegations often surface at the last possible moment, and how foreign actors convert real-world relationships into digital influence. In Taiwan, for example, suspicious online activity may follow physical events, religious exchanges, or changes within local networks. Community leaders and local politicians can therefore serve as early-warning sensors alongside researchers and platforms.
The third lesson is institutional. Election authorities, cyber agencies, journalists, fact-checkers, researchers, platforms, and local leaders must know how to exchange information before a crisis without becoming instruments of government control.
Above all, democracies must act without imitating the systems they oppose. Resilience means moving quickly enough to defend elections, but carefully enough that the defense does not destroy the trust it seeks to preserve.