A cybersecurity CEO says research involving prompting techniques tied to certain Anthropic models that helped lead to U.S. export restrictions was intended for defense and cybersecurity purposes rather than to bypass safeguards. In commentary reported by Yahoo News and Fortune, the CEO characterizes the work as “Defense Oriented Prompting” (DOP), describing it as a way to develop capabilities that protect systems and support defensive teams. The CEO argues the research is not a “jailbreak,” framing it as testing or enabling prompt-related methods that defenders can use to understand risks and improve security. Both outlets reference the same claim and do not attribute additional specific details about the exact technical method, which Anthropic models are covered, or how regulators connected the research to the final export restrictions. The coverage focuses on the CEO’s intent and interpretation of the prompting-related research, positioning it as aligned with defensive needs in the cybersecurity field.