Secure AI Use Without the Blind Spots

The malware known as Latrodectus has become the latest to embrace the widely-used social engineering technique called ClickFix as a distribution vector. “The ClickFix technique is particularly risky because it allows the malware to execute in memory rather than being written to disk,” Expel said in a report shared with The Hacker News. “This removes…

Cisco found and fixed a significant vulnerability in the way Anthropic handles memories, but experts warn that mishandled memory files will continue threaten AI systems.

Researchers detailed a newer double-extortion ransomware group made up of former members of BlackSuit, which was recently disrupted by international law enforcement.

Malicious actors can exploit default configurations in ServiceNow’s Now Assist generative artificial intelligence (AI) platform and leverage its agentic capabilities to conduct prompt injection attacks. The second-order prompt injection, according to AppOmni, makes use of Now Assist’s agent-to-agent discovery to execute unauthorized actions, enabling attackers to copy and exfiltrate sensitive

The financially motivated threat group used cloud resources to conduct a complex, ransomware-style attack against an enterprise victim.