OpenAI's New Agent Aims to Automate Online Tasks, Early Test Shows Bugs

OpenAI has introduced a new always-on agent designed to automate online tasks like buying furniture. In an initial test, the agent was buggy, failed a captcha, and responded in an overly familiar way.

• OpenAI caps paid ChatGPT accounts per device2026-11-01
• First-time voter tests ChatGPT for election info2026-11-01
• Antler, OpenAI weigh Nigeria energy and AI infrastructure investment2026-11-01
• OpenAI may merge ChatGPT and Codex usage limits2026-11-01
• AI smart glasses can answer 80% of SAT questions, study finds2026-11-01