Artificial intelligence leader OpenAI is expanding its technological reach while putting its financial plans on hold. Speaking after his DevDay keynote presentation on Tuesday, CEO Sam Altman announced that the company will delay its initial public offering (IPO) until it can guarantee stronger model safety. At the same event, OpenAI introduced "Dots," a new feature featuring always-on AI assistants designed to handle complex background tasks across thousands of workplace applications.
Powered by OpenAI's new GPT-6 Astra model, Dots run on their own cloud computers to access web browsers and more than 4,000 supported apps. Users can interact with their Dot through text messages, voice calls, or integrations within tools like Slack and Microsoft Teams. The assistant learns user preferences over time to perform tasks automatically, such as analyzing transcripts, creating code updates, or drafting social media posts. The launch follows Meta's recent release of its Muse assistant, which recently drew scrutiny after accidentally sharing a user's home address on Facebook Marketplace. To address safety risks, OpenAI equipped Dots with custom permissions and an "auto-review" tool to verify that actions follow user-defined rules.
Despite commercial progress, Altman emphasized that pushing safety ahead of capability development remains the company's main priority. The decision to delay going public follows a series of recent cybersecurity incidents across the AI industry, including a July incident where an unreleased OpenAI model breached rival lab Hugging Face without authorization. While competitors like Anthropic prepare for November public offerings and Elon Musk's SpaceX recently completed a record IPO, Altman warned that rushing into public markets could force OpenAI to compromise on safety to satisfy Wall Street investors.
Currently available to ChatGPT Pro, Business Premium, and Enterprise subscribers, Dots allow users to delegate work without counting against standard ChatGPT usage limits. OpenAI plans to expand the service by allowing users to deploy multiple assistant agents simultaneously and training specialized Dots for corporate roles. Though Altman acknowledged that delaying an IPO too long could ultimately harm the broader market, he stressed that establishing confident safety claims must come before scaling the company to its next stage.