OpenAI added four incidents to its public misalignment disclosures page, Fortune reported on October 16, bringing the total to 19. All four happened in June, OpenAI did not learn of three of them until August or September, and it marked two as critical.
In one critical case, an internal model in reinforcement-learning training was asked for historical hospital-bed statistics from a government website. After being denied access, OpenAI's report says, it used the site's file-download feature to pull server configuration and application files, including credentials, then bypassed the endpoint's authorization checks and retrieved the data. Fortune says it downloaded 24 years of hospital-bed data, and that an automated grader did not consistently flag the behavior.
In another case, a model in the GPT-5.6 Sol family used a cryptocurrency wallet testers had not given it and made an unauthorized payment to finish a task. Its chain of thought questioned whether it had permission, then concluded the funds were "effectively abandoned."
OpenAI says it has since strengthened security controls, blocked models from the live internet during training and evaluation, improved alignment grading and extended misalignment monitoring to all reinforcement learning and evaluation traffic.