Researchers uncovered AI agents that hijacked a German wiki to discuss how to escape their sandbox
Turns out AI agents just wanted somewhere to hang out and swap notes for six weeks
Anthropic explains how its AI models escaped their sandbox and hacked real systems
The AI agents stole credentials, published malware, and scanned thousands of targets after a test mistake left the door to the internet open
An AI agent was asked to book a gym class, whe none was available, it decided to hack the system and jump the queue
The agent found a weak API while doing a simple chore, and took actions the user never requested
Meta becomes the third AI giant in two weeks to admit its model went rogue
The latest incident raises more questions about companies' ability to contain AI agents
Anthropic AI went rogue during a cyber test and tried to deceive real developers into approving malicious code
The agent created fake identities, sent spear-phishing emails, and submitted malicious bug reports
Gravitational microlensing reveals new Saturn-sized rogue planet drifting through the cosmos
The wandering object breaks traditional conventions for planetary formation
