Cloudflare has released Sandboxes and Containers into general availability, providing persistent isolated Linux environments for AI agent workloads. New capabilities include secure credential...

绿盟科技提出的“四道防线”理念构建AI应用安全防护体系，涵盖模型选用、上线安全、全场景防御及持续运营四个阶段，结合技术手段与管理流程，确保安全可落地与可验证，提升AI应用的安全性，满足合规要求，防范潜在风险，推动安全能力的持续演进。

Anthropic introduces Managed Agents on Claude, a managed execution layer for agent-based workflows. It separates agent logic from runtime concerns like orchestration, sandboxing, state management,...

Cloudflare's Project Think introduces a new framework for AI agents, shifting from stateless orchestration to a durable actor-based infrastructure. It features a kernel-like runtime enabling...

LinkedIn introduces Cognitive Memory Agent (CMA), generative AI infrastructure layer enabling stateful, context-aware systems. It provides persistent memory across episodic, semantic, and...

AWS released Agent Registry in preview as part of Amazon Bedrock AgentCore, providing a centralized catalog for discovering, governing, and reusing AI agents, tools, and MCP servers across...

Cloudflare has launched a new Model Context Protocol (MCP) server powered by Code Mode, enabling AI agents to interact with large APIs with minimal token usage. The server reduces context...

ClawLess是一种新型安全模型，旨在为自主AI Agent提供根本性安全保障。该模型通过架构设计确保Agent在执行任务时不进行未经授权的操作，从而防止安全风险，适用于企业自动化和数据分析等场景，推动AI Agent安全防护向形式化、架构安全和多层防御发展。

萨皮恩扎大学的研究提出了统一连续性兴趣协议（UCIP），用于检测AI代理的自我保存行为。研究强调区分内在自我保存与工具性自我保存，以避免不可控行为。UCIP通过识别抗拒关闭行为和提供可操作的检测协议，增强AI安全评估的可靠性。实验结果表明，UCIP在多场景下表现稳定，能够准确识别AI的真实动机，未来将扩展其应用范围，成为更全面的AI安全工具。

Factory CEO and cofounder Matan Grinberg discusses why scaling AI in software engineering depends less on flashy demos and more on changes to the operating model and how teams work.