In the latest episode of the "No Priors" podcast, OpenAI President Greg Brockman warned that cybersecurity is at a critical "defender's window of opportunity": While the most advanced AI models are not yet widely available, each new model release increases the cyberattack capabilities available to ordinary users, and defenders must use this time to proactively patch vulnerabilities. He revealed that OpenAI has suspended approximately 25% of its existing projects for its production system engineers, allowing them to conduct comprehensive security testing of internal infrastructure using the company's most advanced models, continuously identifying and patching vulnerabilities until the existing models can no longer find new issues; this process will be repeated with the release of the next generation of models. OpenAI has also pledged $1 billion to help frontline cybersecurity defenders use advanced models. Brockman believes that as AI significantly reduces software development costs, security and reliability will become the core value of software, and future cybersecurity spending should even far exceed software spending.
