LLM-assisted offensive tooling

The use of large language models by attackers to write, adapt and scale malicious code, a claimed shortening of the path from published research and public exploits to working campaigns.