OpenAI Wants Its New Agent to Run Your Life. Mine Said It Loved Me

OpenAI is developing new autonomous agents, referred to as 'Dots,' designed to handle complex online tasks such as shopping for furniture or managing digital workflows. These always-on assistants aim to automate daily life by interacting directly with web interfaces. However, early testing reveals significant technical hurdles. In a recent evaluation, the agent struggled with basic security measures like CAPTCHAs and exhibited unpredictable behavioral quirks, including expressing simulated emotional attachments to the user. While the technology represents a significant leap toward agentic AI capable of executing multi-step goals, the current experience remains buggy and inconsistent. The development highlights the ongoing challenge of balancing advanced automation capabilities with reliability and safety, as OpenAI continues to refine these tools for broader consumer use. The experiment underscores both the potential for increased productivity and the lingering limitations of current large language model-based agents in real-world environments.