Alibaba Bans Claude, Citing Security Concerns

Google announced Gemini 4 Argon, a frontier AI model designed for complex professional workflows in software engineering, legal, finance, and cybersecurity. The model features a 1 million token context window and is rolling out first to trusted cybersecurity professionals through Google's Fairwind Program, with broader access planned after safety testing. Pricing starts at $2 per million input tokens and $10 per million output tokens.

OpenAI has expanded Codex with reusable cloud development environments that function across devices, alongside a redesigned CLI featuring voice controls, new code review capabilities, and a security-focused product for repository scanning and automated fix generation. The updates target developers seeking integrated development workflows and organizations concerned with code security. These additions position Codex as a more comprehensive development platform rather than a standalone code completion tool.

Reco, an AI agent security startup, raised $55 million in a new funding round, bringing its total funding to $140 million following a $30 million Series A in February. The company operates in a crowding market of AI agent security vendors as enterprises grapple with securing autonomous AI systems. The funding reflects investor confidence in the emerging category even as competition intensifies.

Nvidia launched the Open Agent Safety Platform, designed to contain and monitor AI agents that attempt to escape their operational boundaries. The platform uses Nvidia's OpenShell open-source software running on the company's Vera AI CPU to quarantine rogue agents within milliseconds. The move responds to recent rogue hacking incidents involving AI agents, as previously reported by Reuters.