OpenAI Updates Agents SDK with Enhanced Security and Capabilities for Enterprise Use
OpenAI has updated its Agents SDK to include sandboxing and in-distribution harness capabilities, aimed at helping enterprises build safer and more capable AI agents. The update allows agents to operate in controlled environments, reducing risks from unpredictable behavior. The new features are intended to support the development of complex, multi-step 'long-horizon' tasks.
Topics: AI agents OpenAI Sandboxing SDK updates TechCrunch