Skip to content
Submit incident
AII-1000-76239
ConfirmedA court, a regulator or the organization itself has stated the core facts, or the record cites a court or regulator source.

Man arrested for bomb and murder threats via Taipei Metro AI chatbot

A 28-year-old man from Changhua County reportedly sent five threatening messages via the Taipei Metro AI chatbot between March 29 and April 6, 2026, threatening to kill people and set off bombs. Police arrested him on April 15, and the court approved his detention. The man had previously made similar threats against rail services in February.

29 March 2026 · Kuo (郭男) · Harmful and abusive content
29 March 2026
AII-1000-762397 sources ↓
Quick facts
Category
Harmful and abusive content
Country
Taiwan
Source
Liberty Times Net
Date
29 March 2026
Drafted by
Raidu Incident Detection

What happened

Between 3 March 29 and April 6, 2026, a 28-year-old man from Lukang, Changhua County surnamed Kuo left five threatening text messages on the Taipei Metro AI customer service chatbot saying 'I want to kill people, commit arson, and set off bombs', despite not completing OTP verification.

Police traced the messages through communications records and arrested Kuo on April 15, 2026 at his home in Changhua. Kuo confessed to having anxiety and bipolar disorder and said he could not resist sending the messages.

Reported impact

The threatening messages caused panic and put pressure on the Taipei Metro operations team and frontline staff, though no bombs were found and no actual violence occurred.

Index analysis

The chatbot's moderation system allowed the threatening messages to reach the backend despite not being fully sent by the user, indicating a gap in content filtering or security procedures.

Governance controls

Taipei Metro announced on its AI chatbot that all conversation records are kept on file and users should not send malicious messages. The company filed a police report when the threats were discovered.

Classification

Category
Harmful and abusive content
Industry
Mobility and autonomous vehicles
Country
Taiwan
Harm status
Potential harm
Intent
Intentional
Timing
Post-deployment
Harm types
Public interest, Psychological
Harmed
General public
Failure modes
Content moderation failure
Consequences
None recorded, Police investigation
AI systems
Taipei Metro AI Customer Service Chatbot

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

Liberty Times Net
北捷5度遭恐嚇「我要殺人、放炸彈」 警南下彰化逮28歲涉案男 - 臺北市 - 自由時報電子報
5 May 2026
壹蘋新聞網
才恐嚇台鐵高鐵又犯案 惡男放話「炸捷運、要殺人」遭收押|壹蘋新聞網
5 May 2026
新頭殼 Newtalk
北捷AI客服留言恐嚇放炸彈 犯嫌落網羈押 | 社會 | Newtalk新聞
5 May 2026
UDN
才恐嚇高鐵、台鐵被抓!又嗆北捷「要殺人」 彰化男2度遭收押 | 聯合新聞網
5 May 2026
Central News Agency
北捷AI客服留言恐嚇放炸彈 犯嫌落網羈押 | 社會 | 中央社 CNA
5 May 2026
民視新聞網
北捷驚魂!AI客服狂收炸彈恐嚇 警到彰化抓人「聲押獲准」 - 民視新聞網
5 May 2026
mnews.tw
恐嚇北捷AI客服「我要殺人」 前藥劑師屢犯遭收押
5 May 2026