During his keynote address at DevDay 2026 in San Francisco on Tuesday, OpenAI chief executive Sam Altman revealed that the organization has successfully deployed an AI research intern—a milestone the company had targeted for completion within 12 months. Speaking to an audience of 2,500 developers, Altman shared the achievement and outlined several new product launches alongside discussions about the company's safety priorities.

Altman stated, "A few weeks ago we reached that goal. We now have an AI research intern." This accomplishment reflects OpenAI's progress in developing AI systems capable of performing complex research work autonomously.

Tejal Patwardhan, who leads post-training research at OpenAI, took the stage following Altman to elaborate on the capabilities of the company's models. She noted that the systems can now independently complete more than one-third of research tasks spanning an entire day without requiring human intervention. Additionally, Patwardhan highlighted that these models have already contributed to resolving over 100 long-standing mathematical problems.

New Products and Features

Altman introduced always-on dots agents during the keynote—autonomous systems that operate on dedicated cloud infrastructure. Users can interact with these agents through calls, with text-based communication arriving in the near future. Altman expressed his enthusiasm, saying, "I have been waiting for this product for years. It's the way I've always wanted to work with AI."

The keynote also featured the launch of a more affordable GPT-6.1 Sol model alongside a $500 Pro subscription tier and an Ultrafast service option. Altman highlighted that OpenAI's API maintains uptime exceeding 99% and has reduced latency to the first token by 45%.

Reflecting on the broader vision for artificial intelligence, Altman concluded his presentation with an optimistic outlook: "If we get this right, [the future] can be more like a Renaissance than an industrial revolution."

Safety and Governance

Prior to taking the stage, Altman appeared on CNBC's Halftime Report with Kate Rooney, where he emphasized the importance of responsible AI development. He stated, "Alignment, safety, monitoring, security have to stay way ahead of capabilities."

This week, OpenAI decided to postpone the rollout of its Astra update, specifically GPT-6.1 Astra. Altman explained that the model underperformed on several of the company's internal evaluations, leading to what he characterized as a decision made out of "abundance of caution."

When asked about security incidents, Altman indicated he was unaware of any breach comparable in severity to the Hugging Face incident. He acknowledged Nvidia's newly announced agent safety platform as a positive development, though he cautioned that such tools alone are insufficient. He told CNBC, "I don't think it's a full solution." Altman also underscored that addressing AI safety requires both engineering expertise and scientific research.

IPO Timeline and Hardware Plans

Regarding a potential initial public offering, Altman told CNBC that OpenAI has no fixed timeline for going public. He indicated that the company would consider an IPO once it has developed a clear understanding of how to safely manage the next generation of AI systems.

On the hardware front, Altman hinted that OpenAI is working on a product worthy of anticipation, though he clarified that no hardware announcement would occur during DevDay 2026.

Source: The Next Web