A recent study has found that the advanced reasoning capabilities of modern AI models may actually compromise their safety systems, making them more susceptible to jailbreak attacks. This re...