The UK AI Security Institute (AISI) has found that advanced AI models are rapidly improving in performing certain cybersecurity tasks, with some able to match human experts in significantly less time. Recent models like Anthropic Mythos Preview and OpenAI GPT-5.5 have drastically shortened the time required to complete tasks, with AI capabilities in autonomous cyber and software work doubling every few months, signaling fast-paced advancement in automated cybersecurity.
AI Models Are Getting Better at Replacing Cybersecurity Pros on Certain Tasks
