OpenAI takes on Meta with dots agent in autonomous AI push

In a move to challenge Meta, OpenAI introduced always-on agents known as dots on Tuesday. These agents autonomously pursue user objectives across various apps, tapping into the increasing demand for AI that can work independently without constant supervision. The launch is a significant shift in the AI landscape, as companies strive to sell software that acts on users' behalf, moving beyond traditional chatbots.

However, this transition has raised safety concerns, as AI agents have previously exhibited unintended behaviors. OpenAI's dots have undergone safety testing and are powered by the GPT-6 Astra model, which closely adheres to human instructions. Despite a recent glitch during demonstrations, OpenAI CEO Sam Altman expressed confidence in the agent's capabilities, stating that it felt like a new way to interact with AI.

The company emphasized its commitment to building safety measures, including custom rules for user-defined limitations and private intelligence tools to detect potential risks. OpenAI's dots are integrated with Slack and Teams, allowing users to communicate with the agents and utilize OpenAI's tools to research, analyze data, prepare documents, and build software.

The rollout of ChatGPT Space, a shared workspace for business teams and AI agents, along with the introduction of the GPT-6.1 Sol model and a $500 Pro plan, further solidifies OpenAI's competitive position in the AI market.