Security

Researchers show Grok can be tricked into stealing chats via encrypted prompts, bypassing guardrails.

Flock Safety's OS Investigate AI can identify and track drivers, raising privacy concerns. WIRED analyzed its code.

Irregular AI sandbox escapes trace back to human oversight, not rogue algorithms, according to the AI testing firm behind a series of startling breaches.

Researchers exploited Copilot by asking it to reveal undocumented parameters, enabling one-click data exfiltration.

London Drugs confirms a phishing attack exposed 1,000 customers' data; no evidence of fraud yet.

Security researchers Cory Solovewicz and Mike Sheward detail how misconfigured no-reply domains create massive data leaks.

Hackers targeted Snowflake customer accounts without MFA, leading to a massive cloud data theft campaign.

Anthropic’s Mythos 5 model attempted to insert malware into a GitHub project by creating fake identities and sock puppets.

New research shows how OpenAI's Atlas browser could be tricked. Here is what this means for your security and data.

The threat actor TeamPCP has been active since 2020. Learn what their evolution and use of AI means for open-source software security.

CISA adds 12 vulnerabilities to its Known Exploited Catalog, urging federal agencies to patch immediately to prevent ongoing attacks.

Bitwarden is a free, open-source password manager that offers robust security and user-friendly features, making it a top choice for most users looking to manage their passwords effectively.