Invest1 distinct publisher3 min readUpdated
Roman Storm says the theory that convicted him would also reach Google and OpenAI. It is a post, not a brief, and the count he was convicted on is the one his analogy fits worst.
The Investor · Invest desk

Compiled by The InvestorSomething wrong?How this is made
Roman Storm, convicted in August 2025 in the Southern District of New York on one count of conspiracy to operate an unlicensed money-transmitting business [3], published a post on August 14, 2026 arguing that the theory behind that conviction would, applied consistently, also reach Google and OpenAI [1]. The argument matters to anyone shipping general-purpose infrastructure because it tests where knowledge of misuse stops being a fact about your users and becomes an element of your offense.
Storm's hook is symmetry. He quoted reporting on North Korean IT workers whose salaries help finance the regime's weapons programs, workers who routinely used OpenAI's ChatGPT for writing, coding and technical assignments and preferred Google's Gemini for image alteration and document forgery [7]. Both companies, he wrote, have long been aware of that misuse and continue to offer the services and collect subscription revenue [8]. Prosecutors in his own case alleged that Tornado Cash, the non-custodial mixer he co-founded [2], facilitated the movement of more than $1 billion in criminal proceeds, including funds linked to North Korean cyber operations [4]. Storm has maintained throughout that he wrote open-source code, never controlled user funds and never directed illicit activity [6].
He then made the reductio operational: issue sweeping subpoenas to the AI companies' employees, scour internal messages for any indication of knowledge of DPRK use, and seek indictments under the International Emergency Economic Powers Act, where exposure runs to twenty years [10]. Jailing those founders is absurd, he wrote, and "if that logic is absurd for Google and OpenAI, it's absurd for Roman Storm. You prosecute the criminal, not the toolmaker" [9]. Twenty years is four times the five-year maximum Storm now faces on the surviving count [14].
The gap between those two numbers is the tell. The jury hung on the money laundering conspiracy and sanctions violation charges and convicted only on the money transmission count [5][3]. The part of the government's case Storm's analogy attacks hardest, the sanctions and facilitation theory, is the part the jury could not agree on, while the count that stuck concerns the business he is alleged to have operated rather than what strangers did with his code [15]. A court asked to rule on the toolmaker question would be reaching past the verdict actually entered [15].
That is also the constraint on the argument: it is a post, not a brief. The reporting describes an August 14 statement in public and records no filing containing it and no enforcement action against either AI company on this theory [16]. The substantive dispute stays open on both sides. Storm's supporters read the partial conviction as a dangerous precedent for open-source and privacy-enhancing software [11]; others argue that developers who design and promote services with clear knowledge of substantial criminal exploitation cross a legal threshold [12]. The Digital Asset Market CLARITY Act would draw the line by statute, protecting code authors from liability for third-party misuse, and its prospects remain uncertain [13].
Three things to watch. Whether Storm's reductio migrates from social media into an appellate filing where a court has to answer it [16]. What happens to the hung counts, since that is where the facilitation theory actually lives [15]. And whether the CLARITY Act's protection for code authors survives, because absent a statute the liability perimeter for neutral infrastructure gets set case by case [13].
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
In a post published on August 14, 2026, Roman Storm argued that the Department of Justice's theory of holding a software developer responsible when third parties misuse a neutral tool should, if applied consistently, also ensnare Google and OpenAI.
Storm is a co-founder of Tornado Cash, a privacy-focused, non-custodial cryptocurrency mixing service.
Storm was convicted in August 2025 of one count of conspiracy to operate an unlicensed money-transmitting business after a multi-week trial in the Southern District of New York, in the case United States v. Roman Storm.
Prosecutors alleged that Tornado Cash facilitated the movement of more than $1 billion in criminal proceeds, including funds linked to North Korean cyber operations.
The jury hung on the more serious charges of money laundering conspiracy and sanctions violations, leaving Storm exposed to a maximum of five years in prison on the remaining count.
Throughout the proceedings and afterward, Storm has maintained that he wrote open-source code, never controlled user funds and never directed any illicit activity.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Single trade-press retelling of a self-published post
One source in the cluster, relaying a defendant's public post and, second-hand, an unnamed investigation into DPRK IT workers. Case facts (conviction, hung counts, five-year maximum, $1B allegation) are specific and internally consistent, but no primary documents, no adverse party comment, and no independent legal analysis are present.
No adoption signal in cluster
The story is a legal-argument story: the supplied source reports no release, deployment, benchmark, pricing or license change, and no enforcement or usage-disclosure event that could be measured as adoption. Inferring uptake of the argument (by courts, regulators, or the AI providers) would require facts the cluster does not contain.
Framing outruns the record
The cluster presents a defendant's post as a test of the DOJ's theory reaching Google and OpenAI, but on the article's own facts the analogy addresses the facilitation and sanctions theory tied to the hung counts rather than the unlicensed money transmission count he was convicted on, and no brief, court ruling, or investigation of the AI firms exists. The underlying legal-precedent question is real, which keeps the gap moderate rather than extreme.
Defendant awaiting sentencing, relayed by industry trade press
The speaker is a convicted defendant facing up to five years and retrial exposure on the hung counts, so he has a direct stake in delegitimizing the theory; his framing also serves the open-source and privacy-tooling constituency that backs him. The sole publisher is a crypto-industry trade outlet whose audience is aligned with that position, and the article carries no adverse comment from DOJ, Google, or OpenAI.
Low - one interested voice, one outlet
The narrow facts (what Storm said, when, and the procedural posture of his case) are reported clearly enough to be reliable, but the cluster's single source, absent adverse comment, third-hand DPRK evidence, and lack of any filing or enforcement action cap confidence in the larger claim that the DOJ theory equally reaches major AI providers.
product
A school agenda shipped with "Vitoiis" and a planet named Marc, and no one read it first1 distinct publisher
product
Google gives US students a $200 AI plan free, plus a renewal date and 5TB of switching cost3 distinct publishers
product
OpenAI shipped a teen ChatGPT. The over-65 cohort doubled to 23% and got nothing.1 distinct publisher
security
The nationalization argument is really a vendor-continuity memo1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.