OpenAI Threatens to Ban Users Who Probe Its ‘Strawberry’ AI Models

image via WIRED

September 17, 2024, 11:00 PM

• •OpenAI's latest AI model, o1, is trained to work through a step-by-step problem-solving process before generating an answer.
• •OpenAI hides the raw chain of thought from users, instead presenting a filtered interpretation created by a second AI model.
• •OpenAI is reportedly coming down hard on any attempts to probe o1's reasoning, even among the merely curious.

OpenAI's latest AI model, o1, is designed to work through a step-by-step problem-solving process before generating an answer. However, OpenAI hides the raw chain of thought from users, instead presenting a filtered interpretation created by a second AI model. This has sparked a race among hackers and red-teamers to try to uncover o1's raw chain of thought using jailbreaking or prompt injection techniques. OpenAI is reportedly coming down hard on any attempts to probe o1's reasoning, even among the merely curious.

Read original article