Security researchers have successfully reproduced Anthropic's Mythos vulnerability findings using off-the-shelf AI models, specifically GPT-5.4 and Claude Opus 4.6. By employing an open-source harness, they managed to replicate the results for under $30 per scan.
This accomplishment underscores the increasing accessibility of powerful AI tools, enabling researchers and security professionals to conduct thorough assessments without significant financial investment. The implications for cybersecurity efforts are profound, suggesting that anyone with access to these AI models can uncover potential vulnerabilities.