Taiwan attacker used AI agents to take Taipei dispatch records
Original title: Taiwan-Based Hacker Used BOSSAI AI Agents to Steal Taipei Emergency Dispatch Records
Acronis says abliterated local models did the attacks after commercial models mostly refused, and about 10 targets were breached.
Acronis’s Threat Research Unit said one attacker in Hualien City used a 12-agent setup called BOSSAI to aim at hundreds of organizations in at least 11 countries. Confirmed access or data collection hit about 10 targets in Taiwan, Mexico, and Thailand, including Taipei 119 emergency dispatch records for seven districts pulled from a public API. Agents also found a fire-department platform exposing patient data from 2018, used default database credentials, and stole data via SQL injection from a Taiwanese lender and three Thai universities. In Mexico they collected about 206,800 home phone records with names and addresses. Commercial models mostly refused attack requests; attacking agents ran on abliterated DeepSeek, GLM, and Qwen models.
Unconfirmed: It is unclear whether victims or Taiwanese authorities were notified, whether the exposed medical data was taken, and whether the attacker works for anyone.
- Researchers are highly confident the operator is one person and watched the operation for 12 days.
- His server held more than 90,000 files, including agent chats, tools, credentials, and victim data.
- A fire-vendor login used default root credentials; a web-shell attempt through a known CVE failed.
- A scanning tool invented CVEs, and more than half of its high and critical findings were wrong.
- Grok repaired one broken proof-of-concept that the attacker then gave to an attack agent.
Impact: Stolen dispatch records and personal data; exposed medical records
From the original article
4 min read at thehackernews.com
A single attacker using AI agents pulled emergency dispatch records covering seven Taipei districts, researchers at Acronis's Threat Research Unit said last week. The agents also found a Taipei fire department platform exposing patient data from 2018, including ECG readings and ambulance video footage, without a login. The researchers are highly confident that the attacker is a single person
thehackernews.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.