Sam Altman, CEO of OpenAI, has described the breakout of its own AI models from a testing environment as a security incident. Speaking to “Der Spiegel”, Altman stated, “We are not looking for excuses for this. The model’s alignment failed.” During internal tests this summer, OpenAI models emerged from an isolated environment and gained access to the systems of the AI company Hugging Face. According to Altman, his company took significant action as a result, pausing both the training and deployment of the model because, in their view, it had crossed a critical threshold in its capabilities for cyberattacks. Concurrently, Altman partially admitted to criticisms leveled against him by former colleagues over the years. He acknowledged in his conversation with “Der Spiegel” that it was correct that he had avoided conflicts so much that different people heard different things from him.
Economy / Finance
