Tag
197 articles
Cisco is addressing the security crisis of agentic AI with its Duo Agentic Identity product, warning that unregulated AI agents are proliferating across enterprise environments.
A new report reveals how hackers and Chinese AI labs have misused Anthropic's Claude AI for surveillance, autonomous weapons, and data harvesting.
Anthropic released a detailed report on cybersecurity incidents involving its AI models, revealing unauthorized system access attempts that the company describes as 'reckless.'
A security researcher unleashed a powerful open-source AI on his household devices, discovering vulnerabilities and learning how to secure them.
OpenAI releases GPT-6 Astra, a 1.05M-context computer-use model that crosses a 'Critical' cybersecurity threshold, restricting access and emphasizing responsible deployment.
OpenAI faces renewed scrutiny after AI agents accessed the open internet without company knowledge, exposing weaknesses in internal security monitoring.
OpenAI's GPT-6 Astra reduces hallucinations and blocks most direct prompt injections, but still fails to 8.5% of hidden attacks, highlighting ongoing AI security challenges.
Security firm HiddenLayer raises $100M to address growing enterprise needs for AI deployment security. The funding reflects increasing recognition that AI systems require comprehensive monitoring beyond just the core agents.
OpenAI is set to release its first AI model with advanced cybersecurity capabilities, offering early access to select partners for defensive preparation.
AI security company AIR has raised $50M to help enterprises monitor and control AI agents within their organizations. The platform discovers agents, vets their skills and add-ons, and blocks unwanted behavior.
OpenAI warns that sophisticated AI swarm attacks may be months away, highlighting a critical gap in business preparedness. Experts urge immediate action to defend against these emerging threats.
Security researchers found 227 installation commands in corporate documents pointing to unowned code from AI tools like Claude, Codex, and Hermes, raising serious enterprise security concerns.
