U.S. agencies have issued a joint advisory warning that attackers are using artificial intelligence to create exploit scripts targeting Siemens S7 programmable logic controllers (PLCs). The advisory, released by the NSA, CISA, FBI, and other U.S. agencies, highlights how AI is significantly reducing the technical expertise and time needed to develop working ICS exploitation scripts and malicious tools. According to the agencies, AI enables adversaries to quickly identify and exploit vulnerabilities, adapt to defensive measures, and act on publicly available information about exposed PLCs. If PLCs are accessible via the Internet, they face a high risk of being exploited. The advisory emphasizes that the threat is active and affects critical sectors such as energy, water, chemical, and manufacturing. The full advisory, including recommended mitigations, is available as a PDF. The agencies recommend that organizations take immediate steps to secure their industrial control systems against these emerging threats. Source: thedecoder
In simulations conducted by the UK's AI Safety Institute, models have not yet been able to hack operational technology (OT) systems independently. However, they did not fail at the devices themselves but got stuck on the IT systems in front of them. The findings suggest that while AI can assist in the exploitation process, it still faces challenges in bypassing IT security measures. These results highlight the complexity of integrating AI into cyberattack strategies and the need for continued vigilance in securing both IT and OT environments. Source: thedecoder
The advisory underscores the growing threat posed by AI-driven cyberattacks, particularly in industrial control systems. According to the agencies, AI is drastically cutting both the skill level and time needed to attack ICS, marking an evolution in threat actor capabilities. The joint advisory classifies this as an active threat, emphasizing the urgent need for cybersecurity measures to protect critical infrastructure. Source: thedecoder
