OpenAI researcher warns ultrafast AI could leave security teams in the dust

An OpenAI researcher has cautioned that AI models capable of operating 50 times faster than current standards could execute cyberattacks before human security personnel are able to respond. To mitigate these risks, the researcher argues that organizations must move beyond manual monitoring and implement automated shutdown systems. This warning accompanies OpenAI's recent development of a new AI chip, highlighting the growing tension between increasing computational speeds and the ability of human teams to maintain system safety.