“Approximately a year and a half ago, ChatGPT was capable of writing effective malware when prompted.”