White House Memo Lets Private Firms Fight AI Cyber Threats

The White House has issued a memorandum on transnational cyber-enabled crime that marks a significant shift in how the United States addresses AI-powered security threats. The policy allows vetted private companies to conduct specified cyber-surveillance operations against foreign criminal organizations, with each operation requiring written government approval.

The framework acknowledges that private companies own and operate most digital infrastructure and cutting-edge technology relevant to national security. Rather than duplicate these capabilities, the government is establishing a model where private technical capacity can be mobilized while authority and accountability remain in public hands.

Why it matters

AI-powered cyber threats operate at machine speed and scale beyond traditional government response capabilities. Autonomous agents can plan, adapt, and replicate across networks faster than agencies can assemble defenses. By pre-authorizing limited private-sector actions, the policy aims to match the speed of AI attacks while maintaining democratic oversight—a balance that will define cyber defense in the AI era.

How AI changes the threat landscape

According to Roberto Baldoni, former Director General of Italy's National Cybersecurity Agency, AI transforms cybersecurity in three fundamental ways: autonomy, speed, and scale. Autonomous agents can plan operations, use tools, adapt after failures, and pursue goals without step-by-step human direction. They operate at machine speed and can be replicated cheaply across borders.

When a private company hosts an AI agent, it may have direct technical control that governments lack—the ability to monitor, restrict, or stop the agent in real time. If an agent behaves unexpectedly or is hijacked, the company can intervene immediately.

Pre-authorized intervention thresholds

The policy establishes that governments should define intervention thresholds in advance. Providers can take narrow, reversible measures on systems they control: revoking access, isolating resources, suspending execution, or placing systems in safe mode. Actions beyond these pre-authorized limits require public authority approval.

A machine-speed offense demands a defense operating at comparable speed. Defensive agents may need to block connections, revoke credentials, or redirect traffic almost instantly. The goal is operational freedom within predefined political boundaries.

The Ukraine precedent

Baldoni points to Ukraine as an example of this model in practice. Hours before Russia's full-scale invasion in 2022, Microsoft detected destructive malware targeting Ukrainian networks, alerted Kyiv, and pushed protections through its Defender service. Ukraine moved critical government data into commercial cloud infrastructure abroad.

This wasn't the traditional wartime model of industry building weapons for the military to operate. Microsoft used commercial capabilities and infrastructure it controlled to support Ukraine's national defense directly. With AI, private firms must not only build defensive tools but operate them.

International coordination needed

AI attacks move through models, cloud infrastructure, and data centers globally. The United States and European allies need compatible protocols for immediate containment, cross-border information sharing, and deployment of defensive agents. While national governments remain the source of legal authority, NATO and the European Union can help align procedures.

Baldoni emphasizes that governments need sufficient technical capacity to judge private-sector assessments. Without this capability, public authority exists only on paper.

The details were first reported by Roberto Baldoni in CEPA's Bandwidth journal.