Taiwan says it was hit by ‘abnormal’ AI-assisted cyber-attack
Taiwan says it detected an “abnormal” overseas cyber-attack against government agencies in July that combined human activity with AI-assisted tools. The case matters because officials describe it as a new form of threat, amid longstanding concerns about cyber operations and other pressure directed at the self-governed island.
Taiwan’s Ministry of Digital Affairs said the attack began on 20 July and has since been investigated and contained. Israeli AI firm Dream reportedly found that at least 85 government accounts were compromised and more than 2,500 personnel records extracted, before the activity spread to the nuclear safety agency and at least seven energy companies; Taiwanese officials did not name China, although reporting said China-linked hackers were suspected. Taiwan recorded an average of 2.63 million daily cyber-attacks on critical infrastructure in 2025, 6% more than the previous year.
- Taiwan says AI-assisted hackers targeted government bodies.
- At least 85 accounts were reportedly compromised.
- Officials did not publicly blame China.
New here? Start with this
Taiwan is a self-governed island that China claims as part of its territory. The two sides have been governed separately since the end of the Chinese civil war in 1949, and Taiwan has its own elected government, military and public services. China has not ruled out using force to bring Taiwan under its control.
Cyber-attacks are one part of the wider pressure Taiwan says it faces, alongside military activity, disinformation and diplomatic isolation. Government departments, energy providers and other essential services are frequent targets because disruption or theft of data can affect national security and daily life. AI tools can help attackers work more quickly or make harmful activity harder to detect, though people are still usually involved in planning and directing such operations.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Taiwan has reason to treat the incident as a serious escalation: compromises of government accounts, personnel records and organisations connected to nuclear safety and energy can threaten public services, privacy and national resilience. The reported combination of human operators with AI-assisted tools may make attacks faster, more adaptive and harder to detect, strengthening the case for urgent investment in cyber defences, information-sharing and preparation across critical infrastructure.
The case against
Officials and the public should be cautious about drawing broad geopolitical conclusions from a single reported incident before full technical evidence and attribution are available. AI may have assisted parts of the operation without representing an entirely new category of threat, and overstating the novelty or assigning blame prematurely could heighten tensions, obscure ordinary security failings and lead to poorly targeted responses.