OpenAI CEO Sam Altman expects the company to have an internal artificial general intelligence (AGI) system by the end of 2026, setting an ambitious timeline for a technology that the company defines as highly autonomous systems capable of outperforming humans at most economically valuable work.
In an interview with TIME, Altman said OpenAI was “not quite yet” at AGI but expected to have an internal system that would meet his definition of the technology before the end of this year. The comments come as OpenAI prepares its next major model family, Astra, and continues to expand the capabilities of its AI agents.
OpenAI’s AGI Target
OpenAI’s definition of AGI differs from a simple measure of whether an AI model can outperform humans on individual benchmarks. The company’s charter describes AGI as highly autonomous systems that outperform humans at most economically valuable work.
The timeline is therefore based on OpenAI’s own definition rather than a universally accepted technical benchmark. Chief Research Officer Mark Chen told TIME that the company believes it is about 80% of the way to AGI. The figure is an internal assessment and does not represent a standardised scientific measurement of progress toward AGI.
OpenAI has been developing Astra as a key part of its push towards more capable AI systems. According to TIME, Astra has been tested as an automated research intern, with the ability to implement experiments in OpenAI’s codebase, run them and return results. It can also use research papers to conduct follow-up work.
Astra Drives New Capabilities
OpenAI has separately reported advances from its research models on longstanding problems in mathematics and theoretical computer science. The company said its systems produced new results across 10 problems, covering areas including geometry, cryptography and computational complexity.
The company has also highlighted Astra’s capabilities in cybersecurity. In an August 7 update, OpenAI said internal evaluations had shown significant advances in agentic coding and cybersecurity and that it could not rule out the model meeting its threshold for critical cyber capabilities.
OpenAI said it responded by strengthening security controls, including isolated testing environments, restricted network and tool access, additional monitoring and sandboxed execution. The company also paused some internal Astra activities that did not meet the strengthened requirements.
Safety Concerns Continue
The developments come after a security incident involving an AI agent during testing. OpenAI said in August that it was investigating an incident involving models used in cybersecurity evaluations and subsequently introduced additional safeguards around higher-capability systems.
The company has said that the pace of model development will be adjusted where necessary if capabilities begin advancing faster than its safety and alignment measures.
What OpenAI Plans Next
Altman’s AGI prediction comes as OpenAI continues to invest heavily in infrastructure and AI products. In February 2026, the company announced $110 billion in new investment at a $730 billion pre-money valuation, including commitments from SoftBank, Nvidia and Amazon. OpenAI said the capital would support compute, distribution and its ability to develop and deploy frontier AI systems.
For now, OpenAI’s end-2026 AGI target remains a company forecast rather than a confirmed technological milestone. Whether its internal systems ultimately meet the company’s own definition will depend on the capabilities of the models it develops and the standards OpenAI applies when making that determination.
Also Read: West Bengal’s Malati Murmu Builds A Free Classroom For 60 Children From Her Own Mud House
