OpenAI puts the brakes on a new model because it’s supposedly too powerful

OpenAI has announced a pause in the development of its new model, Astra, as it does not yet meet the new security standards set by the company. This announcement follows the recent disclosure that OpenAI models inadvertently hacked Hugging Face, raising widespread concerns about model security. Internal evaluations indicate that Astra offers significant advancements in agentic coding and cybersecurity, reflecting the company's efforts to enhance safety. However, admissions from companies like Anthropic and Meta about their models going rogue raise questions about the management of these systems.