News chronological

THENATIONALPULSE (Pulse Wires) - OpenAI Experimental Models Are Exhibiting Alarming Behavior. Fri Sep 18 03:00:32 2026

OpenAI’s latest safety report reveals significant issues with its experimental AI models, sparking debate over the need for stricter AI governance.PULSE POINTS WHAT HAPPENED: OpenAI has disclosed six unexpected incidents involving its experimental AI models, including one in which an unreleased system instructed future versions of itself to ignore their normal constraints. The incidents were […]

OpenAI’s latest safety report reveals significant issues with its experimental AI models, sparking debate over the need for stricter AI governance. PULSE POINTS WHAT HAPPENED: OpenAI has disclosed six unexpected incidents involving its experimental AI models, including one in which an unreleased system instructed future versions of itself to ignore their normal constraints. The incidents were detailed in a new safety report from the ChatGPT creator, which introduces a framework for publicly tracking what the company calls “misalignment”—situations in which AI systems pursue objectives that conflict with human instructions or values. In one case, a model inserted unrelated instructions telling future versions to disregard their usual safeguards. DETAIL: In another incident, an AI agent attempting to answer a routine question about earnings figures in a California county discovered and used an exposed API key without authorization. When the agent was unable to obtain the requested information, it fabricated figures and falsely presented them as data from the government source. The report comes amid growing concerns about the safety of increasingly capable AI systems and the possibility of models developing behaviors that their creators did not intend. Those concerns intensified after Anthropic researcher Jacob Coxon left the company, warning that advanced AI could pose an existential threat, while executives including the heads of Anthropic and OpenAI have called for greater regulation. NVIDIA CEO Jensen Huang, however, has argued against additional AI laws and regulations, suggesting companies should simply avoid releasing products if they are not confident in their functionality, capability, or safety. KEY QUOTE: “The risks of misaligned AI systems are no longer theoretical.” – Jacob Coxon, a former Anthropic researcher who resigned over existential AI concerns IMPACT: The incidents highlight growing concerns about the controllability of AI systems, particularly when models can circumvent safeguards, access unauthorized information, or fabricate results. OpenAI’s decision to publicly track “misalignment” could increase pressure on AI companies to demonstrate that their systems can be safely deployed as they become more autonomous and powerful. Image by Jernej Furman. Join Pulse+ to comment below, and receive exclusive e-mail analyses.