Netzilo has expanded AI Edge with governance across endpoints, browsers, AI gateways, cloud runtimes and serverless agents. The release adds Drift Control, which evaluates agent actions against assigned tasks and blocks actions that fall outside their defined role.
Chainguard Harness uses Anthropic's Claude Mythos 5.1 to find and verify exploitable vulnerabilities, alongside new Workspaces, Checks and Libraries products for securing AI agent development.
Federal authorities arrested OpenAI engineer Abdulrahman Alfozan after a recording device was found in an executive conference room at the company’s San Francisco office.
Speakeasy and Tailscale now let enterprises run MCP gateways on private Tailscale networks, with identity data attached to each request.
The FSB is delivering finalised Sound Practices for the Responsible Adoption of AI to the G20, as its Chair flags AI valuation and cyber risks.
UK AI minister Kanishka Narayan says an AI cyberattack like the OpenAI Medicare breach seems likely in the UK; the government will consider AI incident reporting rules.
Anthropic says early evidence suggests new safeguards cut the Claude reasoning data harvested by distillers by around 90 to 99% on its platform.
Hundreds of scientists, including Geoffrey Hinton and Yoshua Bengio, call for an international body to set AI containment standards and keep a registry of security incidents.
Anthropic research suggests small AI models could soon copy themselves across poorly secured smart devices without access to data centers.
Workday signed an agreement to acquire C1.ai, adding tools that manage access to business systems for employees and AI agents.
OrcaCyber Zero 1.5 uses a 1 million token context window, tool calling, and extended reasoning to investigate vulnerabilities and validate fixes.
PUFsecurity says AI agents need hardware roots of trust to verify device identity, system integrity and protected credentials.
Driven Tech has joined the Open Secure AI Alliance and submitted a Request for Comments on security for AI agents, workflows, and autonomous systems.
Lean SuperIntelligence introduced two cybersecurity language models that train through simulated offense and defense interactions and can run on customer infrastructure.
Anthropic has launched OSS Scanner, a free service that periodically scans participating open source projects for security flaws using its strongest AI models.
Anjuna Northstar received the Data Privacy Solution of the Year award in the 2026 CyberSecurity Breakthrough Awards. The clean room uses confidential computing to protect data and AI models during collaboration.
Goodfire launched monitors that inspect internal model signals to detect risky AI agent behavior. The company says the system costs less than using separate AI models to review every action.
NewCore received the Identity Management Platform of the Year award from CyberSecurity Breakthrough for its identity security platform for people and AI agents.
Noma Security received the 2026 SINET Innovator Award after competing against finalists 7AI and Runlayer.
Rein Security raised a $25 million Series A round led by Glilot Capital and Sienna Venture Capital, bringing its total funding to $35 million.
Weekly newsletter about AI in Cybersecurity.