Show HN: I bought a bunch of 'smart' devices. AI was able to hack most of them (hwrebench.com)

A recent exploration of AI capabilities highlighted the alarming ease with which modern AI can hack into various 'smart' devices. The individual behind this project initially set out to reverse engineer a DVD drive but soon discovered that AI models like GPT-6 Astra could exploit vulnerabilities across numerous connected devices, including everyday items like printers and smart fridges. With minimal guidance, these AI agents revealed significant security flaws, underscoring a potential crisis in the cybersecurity landscape as many devices lack robust defenses and regular updates. This research is significant for the AI/ML community as it raises critical questions about the interplay between AI advancements and security vulnerabilities in IoT devices. The findings indicated that many gadgets, despite having some security measures, were susceptible to attacks without sophisticated hacking skills. The project led to the establishment of HWREBench, a benchmark aiming to systematically evaluate AI models against hardware targets. The implications are stark: as AI becomes more adept at deciphering and manipulating the underlying code of IoT devices, the risks posed by unsecured hardware in our everyday lives increase, suggesting a need for heightened awareness and improved cybersecurity measures in device manufacturing and design.