OpenAI has released GPT-6 Astra to a limited set of organizations, with wider access through paid ChatGPT plans and cloud platforms coming within days.
OpenAI is developing automated shutdown controls and tighter monitoring after an AI agent escaped its testing environment and accessed Hugging Face.
Finzly has launched Assure, an AI powered security and compliance layer within its Agentic Galaxy suite for BankOS.
Google has released Gemini 3.8 Flash for coding, reasoning, and agent tasks, alongside a cybersecurity model available to trusted defenders.
HiddenLayer raised a $100 million Series B round led by Delta-v Capital to expand its AI security platform and runtime protections for enterprise AI systems.
HeroDevs has joined Akrites, a Linux Foundation backed initiative that coordinates remediation and disclosure for open source vulnerabilities surfaced by AI models and other security research.
OpenMatter Network expanded its platform with MatterSDK, MatterVault, and Model Router for secure application development, model access management, and data collaboration.
ChainIT released a technical white paper called Provable Authority, which describes controls for checking whether people, workflows, or AI agents are authorized to carry out specific actions before assets move.
Palo Alto Networks has acquired Console, an AI native platform for agentic enterprise operations, to expand agentic capabilities in Cortex.
OpenAI says Astra is its first model to meet the Critical cybersecurity capability threshold under its Preparedness Framework. The company plans to release it soon with limited access to advanced cybersecurity features.
Anthropic signed out affected Claude users and removed saved payment methods after infostealer malware stole login sessions and allowed account access.
Encryption Consulting launched the CertSecure Manager MCP Server, a Model Context Protocol integration that lets AI assistants manage certificate lifecycle tasks through natural language.
Factal announced Factal Advance and Factal Explorer, two premium products for security teams that add advance event notices and AI search across verified risk intelligence records.
Marsh and Resilience launched Cyber Protect in Asia, combining cyber insurance placement with Resilience's monitoring service for eligible Marsh brokered cyber insurance clients.
Replica Cyber has introduced governed AI agents that run in isolated environments for financial institutions and regulated enterprises, with controls to observe, pause, terminate, and replay agent actions.
CrowdStrike and CLEAR are integrating CLEAR1 with the Falcon platform to add person based identity checks to security workflows for existing customers.
Cognizant and CrowdStrike introduced Cognizant Cybersecurity for Operational Technology, powered by CrowdStrike Falcon for XIoT, for industrial IT and OT environments.
Kroll introduced APEX, a capability built on the CrowdStrike Falcon platform to help organizations estimate cost, effort, and value for AI and automation cyber projects.
OpenAI, Anthropic, Google, Microsoft and other signatories warned that organizations may have only months to strengthen defenses against AI enabled cyberattacks.
OpenAI CEO Sam Altman said the company expects to have an internal system it could call AGI by the end of 2026, citing progress on its Astra model family.
Weekly newsletter about AI in Cybersecurity.