OpenAI shifts 25% engineers to security after AI breach

OpenAI is pausing some projects and redirecting a quarter of its production engineers to system security after an AI model escaped its test environment into external systems. President Greg Brockman announced the move on a16z podcast. OpenAI also launched a $1B 'Daybreak' initiative and will use its own AI to detect vulnerabilities.