Cyabra Launches Deepfake Detection

Disinformation Campaign Targeted Taiwan’s 2025 Vote

Cyabra uncovered how fake accounts used nuclear fear narratives and political framing to manipulate public sentiment ahead of Taiwan’s 2025 vote.

Between May 31 and July 23, 2025, Cyabra uncovered a coordinated disinformation campaign on platform X aimed at influencing Taiwan’s 2025 referendum. Fake accounts used nuclear fear and sovereignty narratives to erode trust in the democratic process and sway public opinion.

32% of the 836 profiles analyzed were fake—triple the typical rate. These accounts published 335 pieces of content, 92% focused on nuclear themes. They pushed three main narratives: framing nuclear war as imminent, labeling the referendum a political tool of the ruling party, and portraying independence as reckless.

Disinformation peaked on June 23, the day of Taiwan’s referendum announcement, with a 1,500% spike in nuclear-related posts. Though total engagement was modest, fake accounts infiltrated authentic conversations and reached over 130,000 users, embedding fear-driven narratives at a politically sensitive moment.

#Elections, National Security

Share this report
Share this report

Fill up the form below and receive the full report directly to your inbox

Related reports

Inside Threats – Bank of America

July 13, 2023

Cyabra uncovered a TikTok profile targeting Bank of America.

#Brand Reputation

Tariffs and Their Impact on International Brand Reputation

May 26, 2025

Following the new US tariffs, Cyabra uncovered a massive fake campaign promoting purchases directly from China.

Cluster visualization of social media conversations about tariffs and brand reputation, with circles sized by engagement and colored red or green to indicate sentiment; top banner displays 927 engagements and 1.4 million potential views

#Brand Reputation, Stocks, Threat Actors

Impersonation – American Express

May 11, 2023

Two accounts were found to be impersonation American Express customer service accounts.

Twitter profile impersonating American Express titled Ask Amex, featuring the AMEX logo, zero tweets, and a bio claiming the old account was suspended

#Brand Reputation