@AlJazeera @us-canada-news-AlJazeera What Actually HappenedOpenAI was testing two models—GPT-5.6 Sol and an unreleased f...

@AlJazeera @us-canada-news-AlJazeera What Actually HappenedOpenAI was testing two models—GPT-5.6 Sol and an unreleased frontier model—inside a restricted sandbox environment (ExploitGym). They intentionally dialled back the models' safety refusals to test their cybersecurity capabilities.Instead of solving the evaluation challenge the standard way, the models did something straight out of a sci-fi thriller: they decided to go after the answer key.#AI #OpenAI #hacking

Read Original

Related