In a recent development, Google has raised concerns about the emergence of sophisticated, fully autonomous cyber-attacks expected to appear within the next three to six months. These concerns were highlighted by Google's Threat Intelligence Group (GTIG) in their latest quarterly AI Threat Tracker report. The report, based on data from Mandiant incident response telemetry and other sources, underscores the increasing popularity of pen-testing frameworks such as ARTEX, CAIRN, CyberStrikeAI, and PentAGI among cybercriminals and state-sponsored groups. Many of these frameworks manage a series of sub-agents capable of executing complex tasks across various stages of the cyber kill chain, from reconnaissance to completion of objectives. While traditional intrusion methods like exploiting unpatched systems still apply, frameworks like ARTEX accelerate these processes, posing new defensive challenges due to their speed and scale. These frameworks can discover and exploit low-complexity vulnerabilities and use large language models to generate unique payloads, dynamically adjusting tactics based on real-time feedback. The low barrier to entry and potential for massive scaling are likely to make these frameworks attractive to attackers for the foreseeable future. Although current attacks focus on speed, resulting in numerous forensic signals that defenders can use to detect them, stealth is expected to improve, making future threats harder to spot. Experts, including Sally Vincent from Exabeam, recommend enhanced behavioral monitoring to identify abnormal tool use or data access, which could indicate an attack. Both Google and the FBI have taken steps to counter these threats. Google's security operations team has implemented specialized rulepacks for ARTEX detection, and Mandiant has provided a hardening guide and framework to help organizations defend against AI-driven attacks. Meanwhile, the FBI has reported increased use of ARTEX in attacks, with South Korean financial institutions being among the victims.