CloudSEK's research reveals that threat actors can manipulate AI summarising tools using prompt injection-based attacks, a technique known as ClickFix.