“Anthropic is working with industry partners, including Amazon, to develop a framework for systematically addressing future government concerns about AI system vulnerabilities.”