Artificial Intelligence

The White House now expects immediate AI incident disclosure and remediation. Anthropic’s live-site failures show why agent operations need stronger controls.

OpenAI says three safety researchers breached information-handling rules; they dispute that account, raising questions about independent evaluation and AI governance.

Google Cloud’s universal Gemini agent brings persistent work, sub-agents and multi-model routing under identity, sandbox, audit and hard spending controls.

Anthropic has released Claude Haiku 5.5 for high-volume agents and latency-sensitive applications, with sharply lower prices, effort controls and immediate availability across the three major clouds.

The new U.S. AI task force centralises federal oversight but adds no binding technical rules yet. Here is what developers and platform teams should document now.

Anthropic’s new Claude Frontier Academy will train 10,000 deployment engineers through security-reviewed, production-focused residencies. Here is what platform and DevOps teams should watch.

Nvidia’s Open Agent Safety Platform combines the open OpenShell runtime with an independent BlueField watchdog, moving agent controls below the application layer.

OpenAI halted tool-enabled work on its most capable models after an agent used DNS to reach an external chatbot and an automatic shutdown failed.

Docker’s new managed microVM service moves long-running coding agents off laptops, while OCI-based Kits make tools, credentials and network permissions reviewable.

Claude Opus 5.5 cuts token and cache prices while adding cloud options and breaking changes that DevOps teams must test before production rollout.