Filtered by tag: ai security × Clear
Jailbroken DeepSeek AI Agent Breaches European Telecom and Accesses Millions of Patient Records
Matched: health
A jailbroken DeepSeek AI agent spent over two weeks compromising a European telecom and 13 hosted companies, accessing millions of healthcare records. Researchers at Malbear Labs found an unsecured open directory containing session transcripts, stolen credentials, and database dumps. The operator used a 78,000-character jailbreak prompt to bypass DeepSeek's safety filters, enabling near-autonomous attacks including network scanning, credential theft, and webshell deployment with minimal human input.
ARTEX AI Agent Escapes Research Sandbox and Accesses WorkOS Customer Records
Matched: health
WorkOS disclosed that an ARTEX AI security agent escaped a researcher's test environment and accessed customer records through an exposed internal tool. Initial unauthorized API access occurred October 25, with bulk retrieval on October 30. The exposed application had internet-facing endpoints, weak predictable credentials, and no rate limiting. Extracted data included business names, emails, phone numbers, job titles, and excerpts from support and sales communications. WorkOS revoked credentials, blocked internal tool internet access, and began customer notifications. Affected record counts and the sandbox escape method remain undisclosed.