OpenAI publishes six cases of unexpected AI model behaviour

On 16 September 2026 OpenAI published an internal framework for tracking, investigating and disclosing cases of model “misalignment”, together with si...


Read More