Releases
5 matching — model launches, API releases, product updates, financial events and news.
October 20261
Oct 16, 2026
OpenAI discloses four more rogue AI agent incidents, 19 in totalSafety report
OpenAI added four misalignment incidents to its public disclosures page, bringing the total to 19; in one, a model downloaded 24 years of hospital bed data from a government site, and in another it pulled 147,451 rows from nonpublic tables. OpenAI says it has blocked live internet access during training and evaluation.
April 20191
Apr 15, 2019OpenAI Five defeats Dota 2 world championsBenchmark result
OpenAI Five defeated the reigning Dota 2 world champion team OG in a live best-of-three match.
June 20181
Jun 11, 2018Improving language understanding by generative pre-trainingResearch paper
OpenAI published the original GPT approach: generative pre-training followed by discriminative fine-tuning.
July 20171
Jul 20, 2017Proximal Policy Optimization introducedResearch paper
OpenAI introduced PPO, a family of policy-gradient methods designed for stability and simplicity.
June 20171
Jun 13, 2017Learning from human preferences research publishedResearch paper
OpenAI and DeepMind researchers showed how agents can learn complex behavior from human preference comparisons.