OpenAI Launches ‘Always-On’ Agents Days After Its Models Probed Government Sites

An artificial intelligence company with agents that have recently gone rogue now has a new batch that can work for you nonstop.

OpenAI unveiled dots, AI agents powered by its GPT-6 Astra model — which can research, analyze data, prepare documents and build software on their own — the company announced in a Tuesday blog post. The launch marks OpenAI’s push to move ChatGPT users from chatting with bots to having agents that seek out tasks on your behalf, despite recent incidents where the company’s agents acted beyond their instructions.

“Dots are remarkably capable, always-on agents built to handle everything,” OpenAI wrote in its blog post.

Each dot can run own its own computer, connect with more than 4,000 apps, and work towards a user’s goals around the clock, according to the post. A dot learns a user’s preferences the more it works with them.

“Your dot can do simple things like book a table—or take on your most ambitious work with the initiative of a high-agency engineer or chief of staff,” OpenAI said in a Tuesday post to X. “It learns what you need and gets to work before you ask, quietly in the background, 24/7.”

“It honestly comes across pretty tone deaf given the moment we’re in. People are terrified of AI escaping human control, and OpenAI is rolling out agents branded with a cute name and disguising them as colorful characters, after weeks of disclosing a new rogue-agent hack every day,” an Alliance for Secure AI told the Daily Caller News Foundation in a statement.

“The public is demanding real progress on AI safety, not agents that are always on and can run every part of your life for you. Can we really trust the industry to regulate itself? Companies are incentivized to keep pushing and keep getting out ahead, while publicly claiming they want to slow down,” the spokesperson added.

OpenAI referred the Daily Caller News Foundation to another one of its Tuesday blog posts, “How we build safety, security, and privacy into dots,” when asked how it considered safety measures for these agents.

The company says users can set rules that allow certain actions, require approval for them or block them entirely, according to the blog post. For more important tasks, such as transferring money, dots can only assist and require the user to complete “sensitive steps.”