Join War on the Rocks and gain access to content trusted by policymakers, military leaders, and strategic thinkers worldwide.
Coverage of the new White House memorandum on combating cybercrime quickly reached the privateer label. CNN called the participants “cyber privateers,” invoking the image of government-sanctioned raiders working the high seas of cyberspace, hauling in digital booty. The reality is considerably less romantic. The administration has tried to create a private offensive cyber program using an exception buried in outdated legislation, and that has shaped how the program must be constructed. For this reason and others, the proposed framework will likely not attract the talent needed to address transnational cybercrime meaningfully.
Within a single month, Congress and the White House each proposed leveraging private cyber capability against foreign attacks, each from different directions. Sen. Mike Lee introduced the Cyber Letters of Marque and Reprisal Act in July, along with a House companion bill from Rep. Tim Burchett. Lee’s pitch is that participants will be “splitting the profits with the federal government” under Congress’s Article I powers to grant “Letters of Marque and Reprisal.” Four weeks later, with Lee’s bill going nowhere, the administration issued a national security presidential memorandum titled “Expanding Capabilities to Combat Transnational Cyber-Enabled Crime,” also focused on combating cybercrimes. However, the memorandum never mentions letters of marque, prize, or Article I, and instead relies on law enforcement authorities.
The memorandum, in effect, creates a contractor program that entails significant operational and legal risk for those who participate. It is based on a reading of the Computer Fraud and Abuse Act’s law enforcement exception that extends it to a private company acting on behalf of the federal government — which, as Jill Goldenziel has noted, has never been tested in court. The uncertain legal foundation is not the only problem. The presidential memorandum offers no apparent financial incentive. It excludes the state-linked actors doing the most damage. It bases operational authority on law enforcement authorities, leaving military and intelligence authorities as coordinating stakeholders.
Free of its privateering imagery, the memorandum reads as a reasonably well-drafted procurement and approval architecture for a program that lets private companies conduct cyber effects operations as agents of the government. It places the program within the National Coordination Center, a body established in January 2025 under Executive Order 14159 and assigned a cyber-enabled crime cell by Executive Order 14390 in March 2026. Notably, the National Coordination Center operates within the Homeland Security Task Force, reinforcing that the program’s organizational placement was driven by the statutory grounds rather than operational design. Two co-executive directors will run the program, one designated by the attorney general and one by the secretary of homeland security, and actions require their approval. Participating companies would undergo vetting, including facilities and personnel security screening, and may be required by contract to maintain a bond or escrow of not less than one million dollars. That bond is forfeitable for non-compliance with that agreement, a term the memorandum conspicuously leaves undefined.
Tom Uren, writing in Lawfare, correctly calls it “unexpectedly measured.” Nothing in it opens a new Wild West of government-sanctioned hackers. Uren is also right that a real capacity gap exists. He argues that the Federal Bureau of Investigation can pursue only the highest-priority groups, that the National Security Agency and U.S. Cyber Command focus on priorities separate from criminal organizations, and that private industry could plausibly close that gap. Nikita Shah and Emily Harding reach a similar conclusion, while warning that the authorities’ question remains unsettled and that the program’s success or failure will likely be based on getting the legal mechanisms right. The FBI’s first unclassified, agency-wide cyber strategy now prioritizes disruption and treats industry partnership as a route to it, and Assistant Director Brett Leatherman wants to move industry involvement from ad hoc cooperation to enduring operational partnerships.
These capacity-framing arguments make a strong case for private sector involvement, but they understate how little the memorandum offers the firms it seeks to recruit. Microsoft’s Digital Crimes Unit has spent 15 years building a disruption capability that never requires the government’s permission. Microsoft’s tactic has been civil litigation rather than direct cyber operations. Using an increasingly standardized playbook, Microsoft sues the operators as unnamed defendants, obtains a sealed court order, and then has the domain registries redirect infrastructure to servers the company controls. The Computer Fraud and Abuse Act is not in play here. Microsoft acts against infrastructure it owns or controls, or through court orders directed at the intermediaries, so no question of unauthorized access arises. Google has built its Threat Disruption Unit using similar methods.
However, civil suits cannot reach infrastructure beyond the jurisdiction of a U.S. court or inside an adversary’s server: deleting tooling, corrupting databases, disabling devices, or recovering data. Those cyber effects operations are all the program would theoretically add, and they sit against everything the company gives up. Target selection passes to two executive directors. Operational tempo becomes a function of government approval. A million dollars goes up against an undefined forfeiture standard. The firm accepts exposure to litigation and employee reprisal. Companies cannot price this many undefined variables. Undefined areas include what triggers forfeiture, liability when an operation goes wrong, the scope of what the government may ask, and the timeline for decisions. Wiley Rein, advising prospective participants, flagged that companies should also evaluate the consequences of declining to perform an approved activity, which means even the option to walk away is unpriced.
That cost-benefit calculation for a potential participant depends heavily on having other means, which Microsoft and Google do. A small firm without its own standing disruption capability likely calculates it differently. The memorandum’s implementing guidance is meant to set eligibility criteria that enable participation by both large companies and smaller, more agile companies suited to specialized tasks: The drafters anticipated two populations, yet only wrote a single set of terms for both. To an established firm, the program is likely a downgrade. For a small offensive shop, it is an excellent point of entry, and its risk appetite is different by construction. This means the program will likely recruit disproportionately from smaller firms, the ones least able to absorb the legal exposure the memorandum leaves unaddressed.
The legal authority underlying the memorandum deserves closer scrutiny. Law enforcement agencies have operated under the Computer Fraud and Abuse Act’s law enforcement exception since 1986, and that much is settled. What is untested is whether the exception reaches a private company acting as an agent of such an agency. In many ways, its operational language reads less as a description of a new policy process than as an effort to satisfy the elements of that exception. The law carries several exceptions, and the one the program reaches for provides that the statute “does not prohibit any lawfully authorized investigative, protective, or intelligence activity of a law enforcement agency of the United States, a State, or a political subdivision of a State, or of an intelligence agency of the United States.” It authorizes participating companies to act as part of lawful investigatory, protective, or intelligence operations carried out by federal law enforcement in accordance with the law. The statute does not address private agents of an agency in either direction. The program’s logic runs from the exception protecting an agency’s activity, to the program belonging to a qualifying agency, to a private company acting as that agency’s agent being covered.
That is a plausible reading. Whether that interpretation holds may turn on how much targeting autonomy the company retains, since a firm executing a government-approved package is easier to characterize as performing the agency’s activity than one pursuing targets the agent selected itself.
A second interpretation of the program also needs to be explored. Counsel at Wiley Rein has suggested that it resolves Computer Fraud and Abuse Act concerns by supplying authorization through a government-approved and supervised process. The difficulty is that the statute does not say where authorization must originate, and the case law has developed around access granted or withheld by the party controlling the system.
The Supreme Court’s most recent treatment of the statute is Van Buren v. United States, and the court held there that a user exceeds authorized access only by obtaining information from areas of a system that are off-limits. Van Buren does not address whether a federal agency may authorize a private company to access a third party’s server. It neither permits nor forecloses it. Criminal infrastructure usually sits on someone else’s cloud instance or compromised hardware. That leaves two plausible interpretations, neither tested in court, and the program commits explicitly to neither.
The memorandum does not impose notification requirements directly on companies. Instead, it states what the implementing procedures must include. What the procedures must include is telling of what the drafters wanted protected, and their priority is hard to defend. One directs procedures requiring a company that discovers its activity has exceeded the parameters of an operation to cease, conduct minimization, and immediately notify the government. The next directs procedures requiring a company that reasonably believes an approved operation may result in loss of life to provide notification. Nothing in it asks the company to stop.
The examples attached to the first procedure are equally telling: covering unintentional targeting of a U.S. person, a system residing in the United States, or a system under the control of a U.S. person. The framework is specific about Americans and mute about everyone else, which matters because criminal infrastructure usually sits on someone else’s cloud instance or compromised hardware. That is the borrowed hardware problem Van Buren does not resolve, now an operational rather than a legal one.
The same logic explains the shape of the approval chain. Wiley Rein noted that the program echoes the review and approval process established for offensive cyber operations in 2018. That memorandum replaced a 2012 directive that had required interagency review and presidential approval, a process critics described as too slow for cyber timelines, and it pushed authority down to the secretary of defense precisely to enable time-sensitive operations at U.S. Cyber Command. The new memorandum runs the other way. It holds approval at two civilian department appointees, with the military and intelligence organizations most experienced against these targets reduced to consultation, because the program is fixed to its law enforcement statute roots.
Historically, a letter of marque was a business proposition, advantageous to all parties. According to the Congressional Research Service, privateers received no government funding and outfitted their own vessels and crews. They were compensated through prize — a large share of the proceeds when a captured vessel and its cargo were condemned and sold — supplemented by bounties paid for specified accomplishments. Congress abolished prize money in domestic law at the beginning of the twentieth century, and no marque legislation has passed since the 1860s, so the machinery would have to be rebuilt. Lee’s bill gestures at rebuilding it. The memorandum keeps only the capital and bond requirements, which leaves an invitation to perform federal law enforcement work at the company’s expense and risk, but only after the bond is posted.
The deeper problem is that the privateering model depends on plunder, and cyber disruption does not reliably produce any. A share of recovered funds is the obvious place to look, and it will not carry the program. Ransomware payments in 2024 totaled roughly $813 million, down from $1.25 billion the year before, and government recovery has been episodic. The showpiece case remains Colonial Pipeline, where the Department of Justice seized $2.3 million in cryptocurrency from a group that had collected something closer to $90 million. Against that modest and unpredictable upside, a participating company would need to maintain continuous capability with facility security, vetted personnel, and the overhead that implies, which could require millions annually.
Because seizures arrive rarely and in amounts too small to fund a standing capability, an appropriation-based model is the only honest answer. A workable structure could pay a company in something like a retainer to hold defined capabilities available and add bounties for specified accomplishments. That is government contracting rather than privateering, and the memorandum concedes the point by making the entire program subject to available appropriations.
Compensation is only half of what a general counsel will weigh. The bond is the smaller exposure. The larger ones are a liability to the owners of infrastructure an operation touches, and exposure under the law of whatever country the targeted servers sit in. Behind both sits the question of whether the United States will stand behind acts it has approved. The memorandum does not say. Its general provisions disclaim creating any legally enforceable right against the government, which does not by itself bar indemnification, since the contracts the memorandum directs could carry it. The asymmetry is the problem. The memorandum has a penalty mechanism aimed at the company, but none at the government. A forfeitable bond sits against conduct it never defines, while nothing at all sits against the harm a compliant firm may cause.
The exclusion of state-linked actors may matter more in practice than anything else in the program, and the distinction fails almost immediately. On paper, the program’s standard appears generous, since a group is presumed not to be operating under foreign government direction unless clear intelligence establishes the connection. That presumption is nearly useless because the boundary it divides has largely dissolved. Russian criminal groups operate with tacit protection and occasional tasking. North Korean operators steal to fund the state. Chinese contractors work for the government by day and for themselves after hours. Scam call centers across Southeast Asia run under the protection of local political elites. Intelligence will almost certainly find that these organizations have some form of state connection, which leaves a target set defined by a distinction that does not describe the threat. Targets surviving that screening enter an approval process that still must be designed, requiring deconfliction across the Departments of State and Defense and the intelligence community before reaching the directors. Any participant can stall a target it has equity in, and none owes the company a decision on any timeline.
Rounding out the concerns is the persistent risk of cyber escalation. The fear is real but misdirected. Uren makes the empirical case by citing WannaCry and NotPetya, which caused enormous global damage yet produced no meaningful consequences for the governments responsible. That establishes something narrower than it first appears. It shows that states have escaped punishment for damage they caused, not that they will absorb disruption of their own proxies without answering. A state is unlikely to lodge a formal complaint that its criminal proxies were disrupted, because doing so would require admitting a relationship it has spent years denying. But the same deniability that forecloses a protest makes a quiet response easy. Shah and Harding put the objection in its sharpest form, noting that a participating firm accepts the possibility of retribution directed at named employees, along with a scope problem that only becomes visible after signing, since a company that joins to fight ransomware crews may find itself asked to work against cartel-linked networks under the narco-terrorism designations. The honest conclusion is not that escalation risk is overweighted, but that it has moved. A government command absorbs retaliation through institutions built for it. A participating company absorbs it through its commercial network, customers, revenue, and staff. The risk does not disappear when the government hands the mission to industry. It shifts to the party least equipped to absorb it, which is perhaps the strongest argument for serious compensation and liability protection rather than leaving either to implementation guidance.
Privateering did not end because zealous or criminal privateers dragged their sponsors into wars. It ended because they could not be controlled. Privateers seized neutral shipping, exceeded the terms of their commissions, and answered to prize courts that were often corrupt and slow. Britain and France agreed in 1854 to stop issuing letters of marque, and the Declaration of Paris made that permanent among its signatories two years later. Private ships armed at private expense, operating for private gain far beyond the reach of any navy, could not be held to their charters. Governance was not the only motive. By the 1850s, Britain had naval supremacy and the world’s largest merchant marine, which made it the power most exposed to commerce raiding and the one with the most to gain from protecting neutral trade. The United States refused to sign for the opposite reason, unwilling to surrender a tool that had twice worked against that same naval supremacy. Both positions follow from strength or its absence rather than principle, and both point toward the same remedy: a system that can enforce its own terms.
So where do we go from here? First, admit that nearly every cyber-criminal organization worth pursuing has some connection to a state, and set the bar somewhere other than affiliation. A conduct test asks whether the target is running criminal infrastructure that victimizes Americans, which is a question the evidence might be able to answer, and leaves state connection as a factor to be managed in planning rather than a gate at the front end.
Second, make participation worth it, and say out loud that the United States accepts some escalation risk in disrupting foreign criminal infrastructure. Shah and Harding argue that deterrent value depends on adversaries understanding what the program is and what it signals, which is impossible if the government treats it as clandestine. But a declaration is cheap when the government making it is not the party bearing the consequence. Congress has the correct tool in its Article I power to grant letters of marque and reprisal and to make rules concerning captures. That remains the aspirational model for fully unleashing the private sector against cybercrime, but it is not a realistic fix. Lee’s bills have little to no prospect of passage: The House companion drew no cosponsors, and no marque legislation has moved since the 1860s.
The achievable legislative action is narrower. First, amend the Computer Fraud and Abuse Act to say what the memorandum can only imply: specifically, that the law enforcement exception applies to private agents operating under written government direction and approval. Second, provide liability protection like that created by the Support Anti-terrorism by Fostering Effective Technologies Act of 2002, which addresses the related problem of tort exposure deterring firms from fielding capabilities the government needs, and which already extends to cybersecurity technologies. The acquisition regulations support this, allowing extraordinary indemnification for any matter that could be designated a qualified anti-terrorism technology. Third, appropriate the money, since the memorandum creates no funding. None of these suggestions are beyond Congress’s ability. Section 1604 of the Senate-reported 2027 National Defense Authorization Act outlines a pilot for contractor-owned, contractor-operated cyber operations under the commander of U.S. Cyber Command. This is the same question of supervised private operators, arriving with a new statutory grant behind it rather than an untested reading of an old one.
The memorandum directs that implementation guidance is due sixty days from signature, in mid-October. Unfortunately, whatever comes out of that process will be written within these constraints, because it cannot fix a statute. The better question is whether Congress and the White House are content with building private sector cyber effects operations capability without purpose-written legislation, or whether a program reverse-engineered from the exception clause of a criminal statute written in the 1980s is sufficient.
Justin Hardy is a career U.S. Navy submarine officer who commanded the ballistic missile submarine USS Alaska and is currently assigned as a federal executive fellow at the U.S. Naval Institute. He previously served as acting deputy director of operations and chief of the Mission Assurance Division at U.S. Cyber Command.
The views expressed are the author’s own and do not reflect the official position of the U.S. Navy, the Department of Defense, or the U.S. government.
Image: Midjourney
