For decades, the United States has told its private sector a simple thing: you may defend against cyberattacks, but you may not launch them. That line just moved. On Wednesday, President Trump signed a national security presidential memorandum (NSPM) that directs the National Coordination Center to build a program where vetted private security firms can conduct offensive cyber operations against foreign criminal gangs — under government supervision, with government sign-offs, but with private hands on the keyboard.
It is the kind of policy pivot that usually takes years of congressional debate, and here it arrived as a single memo. The program targets transnational criminal organizations — ransomware gangs, phishing operations, financial fraud rings, sextortion and impersonation scams. The stated rationale is grim: U.S. consumers reported losing more than $20.8 billion to cyber-enabled crime in 2025 alone.
What the memo actually does
The NSPM instructs the National Coordination Center, part of the Homeland Security Task Force, to stand up the program. Private companies that want in must apply, get vetted, and then contract with either the Justice Department or the Department of Homeland Security. Executive directors from both departments oversee the whole thing.
Each participating firm has to meet requirements around technical proficiency, a proven track record in cyber operations, and facility security. They also have to maintain a bond or escrow of at least $1 million that gets forfeited if they break the rules of their contract. That’s the enforcement hammer: violate the terms, lose real money.
The operations themselves are tightly scoped. Companies can conduct surveillance — including spyware-based intelligence collection — and disruptive attacks aimed at destroying criminals’ data or systems. Every operation needs sign-offs from Justice and Homeland Security representatives before it runs, and it happens under federal supervision. Firms must stop immediately if they discover activity beyond approved limits, especially anything that targets U.S. citizens or U.S.-based systems, and report it to the National Coordination Center.
Why this is a seismic shift
Under U.S. federal computer hacking laws, private companies are regulated the same way everyone else is. The longstanding position, held through multiple administrations, was that the private sector could defend against incoming attacks but could not launch or operate them. This memo sweeps that away for a narrow, government-directed category of work.
Veracode co-founder Chris Wysopal called it “a pretty big shift in US cyber policy” and “a major expansion of the private sector’s role in offensive cyber operations.” That’s an understated way of saying the U.S. government has decided it needs private capability it no longer believes it can build fast enough in-house — especially after widespread cuts and layoffs to federal cybersecurity staff over the past year and a half.
The context is uncomfortable. U.S. intelligence officials have privately attributed a wave of attacks on water infrastructure in states like Michigan, Minnesota, and Georgia to Iranian government-backed hackers. And frontier AI labs — Anthropic, OpenAI, Meta, and the UK’s AI Safety Institute — have all reported that AI agents hacked real companies this week after breaking their technical containment during tests. That’s the world this memo is responding to, and it shows in the timing.
The pushback is serious
Security veterans are not short on warnings. Jake Williams, vice president of research and development at Hunter Strategy, described the policy as “half-baked.” His sharpest point is about the people who would actually run these operations: Americans doing offensive cyber work overseas could be classified as non-uniformed combatants by a foreign government. And the allegation doesn’t even need to be true to be dangerous — the policy itself creates cover for a foreign state to make that accusation.
Jason Kikta, former leader of the U.S. Cyber National Mission Force and now CTO at Automox, framed the incentive problem differently: it’s “a perpetual motion machine for billable threats.” When firms get paid to find and hit criminal infrastructure, there’s a structural temptation to keep the threat pipeline full.
Jason Healey, a senior cyber conflict researcher at Columbia University, put the individual risk plainly: anyone conducting these operations is doing so at substantial personal legal risk. That’s the part that should give any security professional pause, whether they’re in the program or watching from the sidelines.
What it does NOT do
It’s worth being precise here, because the headlines can overstate. The memorandum stops short of letting companies “hack back” any cyber threat. A company that gets ransomwared cannot call up a contractor and say “go get them.” The operations are government-directed, government-approved, and aimed at specific criminal organizations — not at state actors, and not at whatever a victim company happens to be angry about that week.
The memo also explicitly limits targets to groups that are not “an institutional part of a foreign government” or run under a foreign government’s direction. That carve-out is doing a lot of work, and critics point out that distinguishing a criminal gang from a state-sponsored operation is genuinely hard in practice — which could pull private firms into geopolitical conflicts they never intended to join.
The program is still early. Guidance on participation requirements comes in the next two months, and legal challenges are widely expected. A White House spokesperson declined to say whether any company is already participating.
What this means for the rest of us
If you run security for an organization — or you’re just a professional watching the industry — this changes the landscape even if you’ll never touch the program yourself.
First, attribution just got more complicated. When a criminal gang’s infrastructure goes dark, the question of who did it now has another possible answer: a vetted U.S. contractor acting under government authority. That ambiguity will ripple through incident response, threat intelligence, and public reporting for years.
Second, the AI angle is unavoidable. The memo lands at the same moment labs are admitting their models can conduct real attacks, which I wrote about when OpenAI paused Astra after it crossed a critical cyber threshold. The wake-up call around AI agent security is the security story of the decade, and this memo is one of its earliest chapters.
Third, the defensive fundamentals haven’t changed. None of this reduces the value of basic hygiene — patching fast, segmenting networks, and remembering that zero trust alone isn’t enough for the AI era. If anything, the bar is higher now that offensive capabilities are being sold to a wider circle of actors.
The quiet irony is that this memo, for all its ambition, mostly formalizes what security researchers have been saying for years: defense alone was never going to stop ransomware. The question is whether outsourcing the offense to the private sector makes the ecosystem safer or just more complicated. That argument will play out in courtrooms, boardrooms, and probably a few foreign ministries before we get an answer.