OpenAI responded to two prompt‑injection exploits—ShadowLeak and Radware's ZombieAgent—by limiting how ChatGPT handles URLs. The new guardrails restrict the model to opening only exact URLs supplied by users and block automatic appending of characters. While these changes stopped the immediate threats, experts warn that such fixes are temporary and that more fundamental solutions are needed to secure AI assistants. Leer más →

A new Mimecast report finds that cybercriminals are increasingly leveraging generative artificial intelligence to create more convincing phishing, business email compromise (BEC) and multichannel deception campaigns. Phishing now accounts for 77% of attacks, while ClickFix threats have risen fivefold and represent roughly 8% of incidents in the first half of 2025. The report highlights abuse of trusted tools such as DocuSign and Salesforce, and cites the Scattered Spider group as linked to over 900,000 detections. Mimecast recommends multi‑factor authentication, advanced email defenses with anomaly detection, and layered security training to counter the rising AI‑powered threat landscape. Leer más →

Okta kicked off its Oktane conference with a focus on its nonprofit arm, Okta For Good, and the role of AI in security. Guest speaker Jeremy Renner discussed personal resilience and the importance of accessible internet, while Okta executives shared insights on AI agents, threat intelligence, and partnerships with customers such as Zillow. The day blended technical deep‑dives with human‑focused storytelling, underscoring Okta’s commitment to both innovation and community empowerment. Leer más →