Anthropic’s safety philosophy is centered on the principle of “responsible scaling,” which involves prioritizing security over rapid deployment to mitigate catastrophic risks. Their conflict with the U.S. government, particularly regarding lethal autonomous weapons, stems from several core safety mandates:

• Opposition to Full Autonomy: Anthropic firmly rejects the use of its AI in fully autonomous lethal weapon systems. They advocate for a “human-in-the-loop” requirement, arguing that AI should never independently identify, engage, or destroy targets (1:08–1:11).
• Reliability and Security: The company maintains that current frontier AI models are not sufficiently robust or reliable for high-stakes decision-making in warfare, citing a high risk of catastrophic failure and civilian harm.
• Internal Safety Protocols: Through their Responsible Scaling Policy, Anthropic conducts rigorous pre-deployment evaluations designed to prevent their models from being leveraged to create weapons or facilitate mass surveillance.

The Current Dispute: This stance has led to a significant escalation with the Trump administration. Following a government directive on June 12, 2026, Anthropic was ordered to ban all foreign nationals from accessing its latest models, Fable 5 and Mythos 5, due to national security concerns regarding potential “jailbreaks” and safety bypasses (0:56–1:00). Because distinguishing between users in real-time was technically impossible, the company was forced to disable access to both models entirely for all users (1:03–1:05).

Why the US Government Intervened

The Industry Impact