ChatGPT Work: From Chatbot To Autonomous Workplace AI Agent
ChatGPT Work is an autonomous workplace AI agent powered by GPT‑5.6 and Codex that can run multi‑step office tasks across connected apps, files, websites and desktop software for hours, transforming broad user goals into completed documents, spreadsheets, presentations, web apps and ongoing workflows with minimal human intervention. OpenAI introduced ChatGPT Work on July 9, 2026, giving ChatGPT the ability to carry out multistep office work across web, mobile and desktop. This is not another chat interface; it is enterprise AI automation designed to act inside business systems, not just advise from the sidelines. The agent can gather information across apps and workflows to create finished materials like sheets, slides, docs and web apps, then stay with complex projects for hours by breaking them into smaller tasks and completing them independently. That shift towards agentic AI workflows is exactly why security, governance and IT approval cannot be treated as an afterthought.
What The ChatGPT Work Agent Can Actually Do
ChatGPT Work’s pitch is straightforward: take a single goal, hand it to a workplace AI agent, and get back finished work. The agent combines ChatGPT’s language skills with Codex coding capabilities to create documents, spreadsheets, presentations and websites, positioning it as a direct competitor to Claude Cowork in the race for enterprise AI automation. It can pull information from email, calendars, messaging platforms, storage services, CRM systems and project trackers, then use that context to draft analyses, build decks or generate web apps. Users can ask it to take on entire workflows with one request, and Scheduled Tasks allow it to keep projects moving while they are away from their computer or phone. That includes retrieving company information, operating websites and desktop applications, and moving files across systems. In practice, this is AI task automation at a depth that starts to blur the line between assistant and digital coworker.
OpenAI’s Enterprise Play – And The Trust Gap
OpenAI is not hiding its ambitions. ChatGPT Work is its most direct play yet for the enterprise market, launched alongside GPT‑5.6 in three tiers: Sol for complex reasoning, Terra for mainstream enterprise applications, and Luna for high‑volume deployments. Sol is priced at USD 5 (approx. RM23) per million input tokens and USD 30 (approx. RM138) per million output tokens, and OpenAI claims it is 54 percent more token‑efficient on agentic coding tasks than rival models. The company points to security benchmarks as evidence that this workplace AI agent is safe enough for serious work: GPT‑5.6 Sol scored 73.5 percent on ExploitBench, up from 47.9 percent for GPT‑5.5, and is marketed as supporting secure code review, patching and threat modelling. Yet benchmarks are not production reality. As one report notes, security leaders will want to see how those controls perform under real‑world conditions before signing off on autonomous agents touching sensitive business systems. The trust gap is now the central barrier to agentic AI workflows at scale.
What Security And IT Teams Must Lock Down First
If organisations treat ChatGPT Work as a fancy chatbot, they will get burned. IT teams should treat it as an automation platform and start with access, approvals and logging. Because the ChatGPT Work agent can retrieve company information, operate websites and desktop applications, move files and run Scheduled Tasks, every connected system must be inventoried, with clear documentation of whether each connection uses delegated employee credentials, a shared account or a dedicated identity. Access should be limited to the data and functions needed for a defined workflow, addressing the enterprise AI agent security gap created by broad permissions and weak visibility. Approval rules are the next guardrail: administrators can require the agent to seek permission before sending messages, editing shared files, changing calendars or business records, or transferring data. Given that warning labels do not reliably stop users from trusting inaccurate AI output, human review is non‑negotiable when workplace AI agents can act directly on that output.
Governance, Regulation And The Path To Safe Adoption
The launch of ChatGPT Work follows rising enterprise interest in stronger agentic capabilities, but also sharper governance demands. Scheduled Tasks can run once, repeat on a schedule, respond to events or monitor for changes, and Computer Use can click, type and move files across desktop apps and browsers. That power makes workplace AI agents valuable, but it also expands the blast radius of any incorrect instruction or excessive permission. OpenAI’s compliance tooling gives Enterprise and Edu customers logs and metadata that can connect to e‑discovery, data‑loss prevention and SIEM tools, which should be wired into existing audit and incident processes before widespread deployment. External frameworks are emerging too: NIST’s Generative AI Profile provides a voluntary way to document risks, owners, safeguards and testing procedures. Still, security and data governance remain critical barriers in regulated industries, where autonomous agents operating for hours inside core systems will face tougher scrutiny than any previous office tool. Enterprises that want the upside of AI task automation must accept a slower, more deliberate rollout, with security teams in the driver’s seat.
