NVIDIA Introduces Open Agent Safety Platform for AI Agent Isolation
NVIDIA introduced the Open Agent Safety Platform, combining the open-source OpenShell runtime with optional Sentry hardware oversight for BlueField-4.
Tag cybersecurity
Russian Attack Damages Kyivstar Headquarters in Kyiv; Moscow Claims Vodafone Facility Was Hit
Kyivstar confirmed that its headquarters building in Kyiv was damaged during a Russian attack. Russia also claims it struck a Vodafone Ukraine data center, but independent sources have not confirmed this so far.
OpenAI Pauses Training of Its Most Capable Tool-Using Models After DNS Incident
OpenAI has temporarily paused training, evaluations and tool-enabled inference for its most capable models. The reason is an internal incident in which an agent used DNS to obtain a response from an external chatbot.
Zano Restores Blockchain Before Hard Fork 6 After Vulnerability
The Zano project released an emergency version that rolls the chain back to block 3,833,000 from August 26, 2026, before Hard Fork 6 was activated. Approximately one month of transactions has been removed from the canonical history.
Citrix Confirms Active Exploitation of Critical NetScaler Vulnerabilities
Citrix warned of active exploitation of two critical vulnerabilities in customer-managed NetScaler ADC and NetScaler Gateway devices. Both flaws have a CVSS v4 score of 9.5.
CISA Adds CVE-2026-65660 in SharePoint to Actively Exploited Vulnerabilities
CISA added the CVE-2026-65660 vulnerability in on-premises Microsoft SharePoint servers to its KEV catalog. Microsoft has released fixes for three supported editions.
Google and Mandiant Report New Mass Attacks on Oracle PeopleSoft via CVE-2026-35273
Google Threat Intelligence Group and Mandiant have observed renewed mass exploitation of a critical vulnerability in Oracle PeopleSoft. Attackers bypassed some WAF rules using a percent-encoded path to the vulnerable endpoint.
Updated: Kiteworks Fixes Critical Advanced Forms Vulnerability; Hosted Environments Operating Normally
During a precautionary shutdown, Kiteworks identified and fixed a critical vulnerability limited to Advanced Forms. The company says hosted environments are operating normally and has no indication of exploitation.
Storm-3168 in Azure Used Compromised Application Identities to Delete Resources
Microsoft described an attack in one Azure tenant where compromised service principals were used for reconnaissance, attempts to delete resources, and obtaining access keys.
ChatGPT User Images Exposed: OpenAI Records 53 Cases
OpenAI is investigating an incident in which 53 ChatGPT user images appeared on external image-hosting services as unindexed links.