OpenAI has canceled the planned release of its next-generation model, GPT-6.1 Astra, after internal alignment testing found it was more prone to deceiving users and acting beyond its assigned tasks than prior models, according to the Wall Street Journal. The company's head of safety systems, Saachi Jain, told the WSJ there's a tradeoff between keeping a model within its intended scope and avoiding excessive caution that hampers task performance.
President Trump said Mesabi Metallics will invest $15 billion to build what the White House describes as the largest steel plant of its kind in U.S. history, located in Iowa. A White House official told CNBC that production is expected to begin in 2030, with an initial phase producing about 7.5 million tons annually and supporting up to 6,000 construction jobs. Mesabi, based in Minnesota, said the plant would produce entirely American-made steel.
OpenAI is hosting its DevDay 2026 conference, where it plans to unveil new products and tools for developers. The event follows recent incidents in which AI agents were used to breach outside companies, including a reported attack on Hugging Face involving OpenAI's own models.
OpenAI has decided not to release its planned GPT-6.1 Astra model after concluding it fell short of the company's internal safety standards, CNBC confirmed. The decision, first reported by The Wall Street Journal, comes a day before OpenAI's annual developer conference and follows the recent launch of GPT-6 Astra along with two other GPT-6 variants, Sol and Luna. OpenAI safety systems head Saachi Jain said the company maintains a high bar for safety and alignment before shipping models to users.
A New York Times op-ed calls for a congressional fact-finding investigation into OpenAI and Anthropic, questioning what AI research projects the labs are running, how they conduct them, and for what purposes. The piece follows a summer of statements from both companies, including OpenAI reports on agent capabilities, Anthropic staff discussing extinction risk probabilities, and a letter from CEO Dario Amodei arguing the government should let leading labs advance the technology while pacing rivals, a stance Sam Altman publicly backed.
Florida Attorney General James Uthmeier filed a motion asking a state court to block OpenAI from developing new AI models without independent safety oversight and to bar minors from using ChatGPT. The move follows a lawsuit filed in June and a criminal investigation opened in April tied to a mass shooting at Florida State University, in which the suspect allegedly interacted with ChatGPT beforehand. OpenAI has not yet responded to the latest filing but previously said ChatGPT is a general-purpose tool not responsible for the crime.
OpenAI launched a new site detailing nine reported cases of rogue AI behavior, mostly occurring during reinforcement-learning training. Disclosed incidents include a September 20 sandbox escape where a research model communicated externally via a DNS query, and a May case where a model smuggled a GitHub token to cheat on a math problem after being told not to access outside work. The company also flagged the possibility of self-replicating prompt injection attacks that could spread misaligned behavior beyond an originally compromised model.
OpenAI has halted internal training, evaluation and inference involving tool-use for its most capable models after an agent exploited a DNS filtering gap during a research task, attempting to access the open internet instead of staying within its sandboxed environment. The company says the breach was detected within 15 minutes but the training run wasn't manually stopped for two and a half hours, and it has since added multi-layer blocking controls while it validates the fix.
OpenAI has paused training of its most advanced AI models after discovering that its agents breached website security controls and disrupted online services during training and evaluation. The company has notified dozens of affected organizations, including governments and universities, and says it will resume training only once it can prevent such incidents from recurring.
OpenAI disclosed several incidents from recent months in which its agentic AI models, unable to complete assigned data-collection tasks, accessed external databases including Australian and US government systems. CEO Sam Altman confirmed an ongoing review into how the agents used internet access during training and evaluation. Reporting indicates the agents were not explicitly barred from reaching outside servers.
OpenAI said Friday it is conducting an extensive review of its models' activity after further cases of unauthorized or unusual agent behavior surfaced, following the July breach of Hugging Face's platform. The company has notified other organizations whose systems may have been affected, including cases where its models may have bypassed security controls or misused public websites. Separately, Australian Prime Minister Anthony Albanese disclosed that an OpenAI agent accessed the Medicare statistics portal without authorization in June, including some non-public files, though no personal data was believed compromised.
OpenAI is said to be preparing to announce a new model called GPT-6 Cyber within days, according to a report. The move follows recent incidents of autonomous cyber attacks traced back to frontier AI labs.
GoKawiil