Former security researcher testifies that OpenAI's security team was laid off, Microsoft bypassed reviews to launch GPT-4

CryptoWorld News reports that former OpenAI security researcher Rosie Campbell testified that after OpenAI gradually shifted toward a product-oriented approach, it disbanded two dedicated security teams. She mentioned that Microsoft had launched an unrevised version of GPT-4 through Bing in India without review by the internal security committee. Campbell stated that although the model itself posed little risk, as technology advanced, it was necessary to ensure the reliable execution of safety procedures. Former director Tasha McCauley’s testimony pointed out that Altman lied to the board before releasing GPT-4 Turbo, claiming that the legal department had already determined that review was unnecessary. She said Altman fostered a culture of lying, leading to frequent management crises. The last witness from Musk’s side was David Schizer, former dean of Columbia Law School, who stated that Altman’s behavior was almost entirely inconsistent with the practices of nonprofit organizations. The probability of Musk winning on Kalshi has dropped from 40% to 36%.