Tag: secure model development

OpenAI Halts Astra Model Tests Over Advanced Cyber Capabilities

OpenAI is hitting the pause button on internal Astra activities that don't meet its new, stricter security control requirements, following concerns over the model's advanced cyber capabilities. The company is implementing enhanced safeguards, including isolated testing environments and encryption, to ensure responsible development.