OpenAI adds workspace agents

- OpenAI introduced 'workspace agents' in ChatGPT for teams to automate complex tasks and team workflows. - Agents can execute cloud tasks, be shared across teams, include admin controls, and are free until May 6th. - OpenAI also released an open-source Privacy Filter to scrub sensitive data before users paste it into chatbots. ( )

OpenAI has started rolling out workspace agents in ChatGPT, giving Business and Enterprise teams shared AI workers that can run multi-step jobs in the cloud. (openai.com) OpenAI said the agents launched April 22, 2026 and are “powered by Codex,” the company’s coding and task-execution system. The rollout is gradual over the next few weeks for ChatGPT Business and Enterprise workspaces. (openai.com; help.openai.com) These agents are built for repeatable work: preparing reports, writing code, responding to messages, and moving information across connected apps. OpenAI said teams can use them inside ChatGPT or in Slack, and connect tools such as Google Drive, Google Calendar, Slack, and Microsoft SharePoint. (openai.com; openai.com; help.openai.com) OpenAI is positioning the product as a step beyond custom GPTs, which answered prompts but did not persist as shared team workflows. In its launch post, the company called workspace agents “an evolution of GPTs” that can be built once, shared across an organization, and improved over time. (openai.com) The company also wrapped the launch in workplace controls. OpenAI said agents operate within an organization’s permissions, can be previewed before publishing, can run on a schedule, and support governance features including admin controls. (openai.com; help.openai.com; openai.com) OpenAI tied the new feature to a broader push to make ChatGPT a system of record for office work, not just a chat window. The company spent the past year adding connectors for Gmail, Outlook, Microsoft Teams, SharePoint, GitHub, Dropbox, and Box, plus shared projects for teams. (openai.com) At the same time, OpenAI released a separate privacy product aimed at a problem that slows workplace adoption: people paste sensitive text into chatbots. On April 22, 2026, the company published OpenAI Privacy Filter, an open-weight model that detects and redacts personally identifiable information before text is shared. (openai.com) OpenAI said the Privacy Filter can mask names, addresses, phone numbers, email addresses, Social Security numbers, and other identifiers in unstructured text. The model card says it is a token-classification model built for high-throughput privacy workflows and long documents, with a 128,000-token context window. (openai.com; cdn.openai.com; community.openai.com) The two releases land as OpenAI pushes deeper into paid workplace software, where buyers care less about one-off answers than audit trails, permissions, and automation across company systems. For now, the company is using a free trial to seed adoption: OpenAI said workspace agents are free for Business and Enterprise workspaces until May 6, 2026. (openai.com; openai.com)

Get your own daily briefing

Scout delivers personalized news, insights, and conversations tailored to your role and industry.

Download on the App Store

Shared from Scout - Be the smartest in the room.