"This committee has been briefed extensively on our latest AI capabilities, Anthropic in particular, the Mythos model, and the amount of destruction that Mythos can do if it’s in the wrong hands," Rep. Michael McCaul said, setting the tone for a tabletop that forced lawmakers to confront how artificial intelligence could reshape a future China–Taiwan crisis.
CSIS tabletop exercise: lawmakers put through a Taiwan crisis
The Center for Strategic and International Studies convened a simulation that brought members and staffers from the House Homeland Security Committee and the House China Select Committee into a compressed, high-pressure decision environment. Participants included Republican Reps. Eli Crane of Arizona, Michael Guest of Mississippi and Michael McCaul of Texas, along with Democratic Rep. Shri Thanedar of Michigan and Puerto Rico Resident Commissioner Pablo José Hernández.
The event aimed to show lawmakers the trade-offs and uncertainties U.S. officials would face if a fast-moving conflict involving China spilled into American cyberspace, concentrating on the networks that would support the movement of forces and supplies.
The scenario: summer 2027, a maritime quarantine and hacked logistics
Exercise controllers opened with a specific scenario: China launches a massive military exercise around Taiwan in the summer of 2027. U.S. analysts in the simulation must decide whether Beijing’s actions are coercive, amount to a de facto blockade, or signal preparations for a full invasion. The readout stresses that Beijing avoids using the term “blockade,” though the moves effectively quarantine Taiwan.
Simulated U.S. intelligence in the exercise also flags widespread targeting of U.S. transport and logistics systems: ports, freight rail, airports and other transportation networks that would be used to move American forces and supplies. Some simulated intrusions produce immediate disruption; others appear designed to preserve access for later leverage.
AI-enabled tactics used in the simulation
Organizers introduced a series of AI-linked attack vectors. Scenarios included the use of stolen credentials to penetrate systems; interference with logistics and routing systems that control movement of goods and cargo; an AI-assisted misinformation campaign; and a final sequence involving a prompt injection designed to trick an AI into following malicious instructions rather than its safeguards.
Lawmakers were reminded explicitly of recent developments in cyber-capable AI: McCaul referred to the Mythos model, describing it as a powerful, cyber-focused AI unveiled in early April and warning that "we know China is three to five months behind us." Earlier in the exercise he observed, "Cyber is always going to be a leader in any kinetic war."
Tough operational choices: what to defend, who to tell, when to move forces
The war game forced participants into three core operational dilemmas. First, they had to prioritize which networks to defend immediately when multiple critical nodes were under attack. Second, they weighed how closely to coordinate with private-sector operators of ports, railways and airports—systems that in the U.S. are largely not government-run. Third, participants confronted whether moving U.S. forces and supplies would deter Beijing or risk escalation.
The simulation underscored that some intrusions seemed aimed not to destroy systems straightaway but to maintain covert access that could be exploited at a later point, changing the calculus for both public disclosure and active remediation.
What this means for technologists, policymakers, and private infrastructure operators
- Technologists and security teams: The scenarios spotlight automated and AI-augmented tactics—stolen credentials, routing manipulation, prompt injection—that will demand rapid detection, layered authentication, and careful scrutiny of AI outputs. The exercise portrayed these techniques as capable of producing both immediate outages and latent access.
- Policymakers and regulators: Lawmakers in the room were asked to balance military signaling against escalation risks and to consider when to share sensitive intelligence with private operators. The simulation presented a direct policy tension: protecting operational secrecy versus enabling rapid private-sector mitigation.
- Private-sector critical infrastructure operators: The scenario placed these operators at the center of national defense logistics and highlighted the practical dilemma of coordinating with government agencies while defending commercial systems that control ports, rail and airports used for deployments.
The tabletop did more than dramatize a cyberthreat; it mapped specific trade-offs that would arrive quickly in a real crisis. It also tied contemporary worry about advanced hacking groups—referenced by organizers in the context of long-term operations such as Volt Typhoon—to the accelerating capabilities of AI-enabled tools. U.S. intelligence, the simulation noted, assessed earlier this year that China is not planning to invade Taiwan in 2027 and believes Beijing still prefers unification without military action—yet the exercise made clear how rapidly cyber and AI tools could complicate that assessment.
For lawmakers and operators who took part, the takeaways were concrete and sobering: AI can multiply the speed and reach of attackers; private-sector systems are potential pressure points in conflict; and decisions about disclosure, defense prioritization and force movement carry immediate operational and political costs. The next step, in the words framed by the simulation, is turning those trade-offs into policies and procedures that would work under the compressed timelines of a real crisis.
