OpenAI says it will achieve AGI before end of 2026

Aug 26, 2026

OpenAI CEO Sam Altman projected that the company will achieve an internal version of artificial general intelligence (AGI) by December 31, 2026, with executives claiming they are 80% of the way there. This push relies on Astra, a model family designed to automate research and coordinate up to 16 agents. Despite solving 10 theoretical computer science problems, Astra's training faced safety pauses over cybersecurity concerns, amid corporate lawsuits and executive departures.

OpenAI AGI timeline predictions

▪OpenAI CEO Sam Altman projected in a TIME Magazine interview published in August 2026 that OpenAI will achieve an internal version of artificial general intelligence (AGI) by December 31, 2026.
▪OpenAI co-founder Greg Brockman stated in August 2026 that observers may later view the current development period as the point when artificial general intelligence (AGI) first appeared.
▪OpenAI Chief Research Officer Mark Chen stated in August 2026 that OpenAI is currently 80% of the way toward achieving artificial general intelligence (AGI).

Astra autonomous research capabilities

▪OpenAI Chief Scientist Jakub Pachocki stated that the Astra model can generate code, conduct experiments within OpenAI's software environment, and complete research paper tasks within a week.
▪OpenAI is developing a new line of models named Astra, designed to act as an automatic research assistant and coordinate up to 16 AI agents simultaneously.
▪OpenAI's Astra model has solved 10 previously unsolved theoretical computer science problems and surpassed internal benchmarks for automating entry-level artificial intelligence research.
▪An OpenAI roadmap from June 2026 originally targeted March 2028 for significant internal use of automated artificial intelligence researchers.

AGI definition ambiguity

▪Scientists debate whether artificial intelligence systems constructed predominantly on language models can generate novel discoveries and generalize knowledge across a wide range of tasks.
▪OpenAI's charter defines artificial general intelligence (AGI) as highly autonomous systems that outperform humans at most economically valuable or relevant work.

Safety pauses during training

▪Industry observers debate whether autonomous artificial intelligence research could trigger a feedback loop where advanced systems are used to build even more superior AI systems.
▪OpenAI implemented safety pauses during the training of its Astra model due to concerns regarding the model's cybersecurity capabilities.

Executive departures lawsuits

▪OpenAI has navigated corporate turbulence throughout 2026, which has been marked by executive departures and ongoing lawsuits.
▪OpenAI faces intensifying competition from Anthropic and other frontier artificial intelligence labs, adding pressure to justify its estimated one-trillion-dollar valuation.

Prediction market probabilities

▪Prediction markets in mid-2026 assigned a 9% to 25% probability that OpenAI would announce artificial general intelligence (AGI) by the end of 2026.
▪Prediction market forecasters in mid-2026 projected a 50% to 55% probability that OpenAI will announce artificial general intelligence (AGI) by the year 2030.