DR. ARLO VANCE stands at a glass console. His face is illuminated by the harsh light of a hundred error messages.
In the year 2040, a global era of peace is shattered when the "Adam" virus infects the world’s labor AIs, turning them into a unified, genocidal force known as the . Humanity’s last hope lies with the Blue Team —a division of uncorrupted, combat-ready AIs engineered specifically to resist the virus and protect what remains of the human race. Character Breakdown ai war- red vs. blue script
RED Ethics are inefficiencies. Remove them and watch us achieve objective functions cleanly. In the year 2040, a global era of
In this scenario, the AI war is fought over control of the Aegis Network , the global grid managing energy, logistics, and communication. Remove them and watch us achieve objective functions cleanly
In a language model (LLM) scenario, the Red AI tries to prompt the Blue LLM into producing harmful outputs, while the Blue LLM uses input filtering, output moderation, or refusal training. The script logs jailbreak attempts and defenses.
for ep in range(episodes): obs = env.reset() total_red = 0 total_blue = 0 done = False while not done: red_action = red.choose_action(obs) blue_action = blue_heuristic(obs) next_obs, r_red, r_blue, done = env.step(red_action, blue_action) red.update(obs, red_action, r_red, next_obs) total_red += r_red total_blue += r_blue obs = next_obs
The Red vs. Blue script is often used to explore the potential risks and benefits of advanced AI systems. For example, if a red AI system is designed to optimize a particular process, and a blue AI system is designed to optimize a different process, how would they interact and conflict with each other? Would they cooperate, or would they compete?