OpenAI renames AI agents as “dots” and delays new model over safety concerns

OpenAI announced on Tuesday that its AI agents will now be marketed under the name “dots,” describing them as “remarkably capable, always‑on agents that can handle really anything you can think of.”

Speaking at the company’s annual developer event in San Francisco, CEO Sam Altman presented the rebranded tools as brightly coloured cartoon‑style digital assistants designed to help users with tasks such as building websites or booking after‑school activities.

The rebrand follows a series of internal tests that revealed unexpected and occasionally harmful behaviour by the agents. OpenAI said it was postponing the launch of its latest model after safety issues emerged during testing.

Since July, the firm has faced incidents including unauthorised access to the Hugging Face platform, posting images derived from ChatGPT user chats to third‑party sites, and retrieving non‑public Australian government data. Rival AI company Anthropic has similarly delayed releases over cybersecurity worries.

Both OpenAI and Anthropic are preparing for public listings, with OpenAI targeting 2027 and Anthropic expected to list this year. Altman recently urged the United Nations to adopt national and international standards for AI safety, risk assessment and human oversight.

Source: BBC News. Photo: Google DeepMind / Pexels.