OpenAI has announced Dots, proactive AI agents designed to work autonomously on tasks for users, completing work without constant supervision. The company unveiled the new agents at its DevDay event in San Francisco, positioning them as always-on systems that can anticipate users' needs and handle routine tasks across thousands of connected applications. This launch demonstrates OpenAI's continued investment in AI agents, despite recent security concerns surrounding such systems.

Dots can access over 4,000 apps through OpenAI's plugin ecosystem and integrate with Slack and Microsoft Teams, allowing users to interact with them across multiple platforms. At launch, Pro, Business, Premium and Enterprise users in select markets will receive one free primary Dot that can be customised with a name; OpenAI plans to eventually enable users to control multiple Dots simultaneously. The company has emphasised safety features including password protection without model exposure, read-only restrictions on proactive tasks, and customisable approval rules, addressing concerns about autonomous AI systems' security and reliability.