The Export Control That Fits Inside a Passport
American export-control law has held for decades that you can export something without shipping it. Under the rule called deemed export, releasing controlled technology or source code to a foreign person inside the United States counts as an export to that person’s home country. A postdoc reads a drawing in a laboratory in Cambridge or Pasadena, and in the eyes of the regulation a border has been crossed. It was always a slightly uncomfortable idea, because it made a colleague into a jurisdiction.
What kept it from swallowing academic science is the fundamental-research exclusion — work ordinarily published and shared broadly — alongside published information and certain immigration statuses. Universities have lived inside that carve-out for a generation. It is the carve-out, more than the rule, that is under pressure this summer.
Two governments, working from opposite premises, spent three weeks moving the working object of technology control away from the thing being shipped and toward affiliations, collaborations and the freedom to travel. Neither measure is written for quantum. Quantum is simply where the change is easiest to watch, and the reason is arithmetic.
Two documents, one notification, two deadlines
31 July 2026. China’s State Council publishes the Exit and Entry Administration Provisions: nineteen articles, effective 15 September 2026, and in DLA Piper’s reading one of the most significant measures in this area since the Exit-Entry Administration Law of 2013. Most of it is procedural. One provision is not. Where a Chinese citizen violates export control or technology import and export administration regulations and may endanger national industrial or technological security, competent State Council departments may decide to prohibit that person from leaving China.
17 August 2026, first document. The Office of Science and Technology Policy publishes the National Security Science and Technology Strategy. Its list of critical and emerging technologies runs to fourteen entries, among them quantum information technologies. The strategy holds up, as a model of what field-specific protection can look like, the FBI-led Quantum Information Science and Technology Counterintelligence Protection Team — an interagency group pairing quantum specialists with security professionals and pointed at the research community itself.
17 August 2026, second document. The same day, the Department of War announces that the Office of the Under Secretary of War for Research and Engineering has notified thirty American universities and specialized technology institutes to audit active institutional ties to foreign entities identified under Section 1286 of the fiscal 2019 National Defense Authorization Act, and to organizations associated with rebranded Confucius Institutes. They must assess the exposure of sensitive or export-controlled research and implement mitigation plans, which may include termination of problematic partnerships. The fiscal 2025 version of the Section 1286 list carries roughly 130 organizations in China, Russia and Iran. The announcement is public; the individual notifications are not, and the Department did not name the thirty institutions. A US official supplied that list to DefenseScoop.
31 August 2026. Findings and actions are due back, with future federal research funding as the stake.
15 September 2026. The Chinese provisions take effect.
The asymmetry inside the Chinese text
In DLA Piper’s reading of the provisions for multinational employers, the load-bearing words in the exit-ban trigger are may endanger: a forward-looking standard requiring no completed harm, no prosecution, and no finding that anything was actually transferred. The same analysis flags a second feature that is easy to miss. This category of ban carries no fixed time limit, where other categories in the same instrument run six months to three years — a potentially open-ended consequence, attached to an individual, on a standard of possible endangerment.
The American action works through a different lever and does not reach nearly as far. It bars no one from travelling and imposes nothing on an individual. It conditions an institution’s access to federal research money on that institution reviewing, and where warranted unwinding, its ties to listed foreign entities — an audit performed by the university, from its own records, against a list it did not write. Nothing in the published record suggests ordinary international collaboration is the target: the Department’s own decision matrix for fundamental-research risk treats co-authorship and patents, on their own, as insufficient to deny funding. The reach is over partnerships and money, not over people.
Why quantum is the legible case
Fields differ in how legible they are to this kind of control, and quantum information science is unusually legible. The number of laboratories doing serious work is countable, papers carry long cross-border author lists, and the people who can operate the equipment trained in a small number of groups, often in more than one country. When the FBI stands up a counterintelligence team scoped to a single scientific discipline, it is because the discipline is small enough for that to be a coherent unit of work.
That legibility cuts both ways. Korea’s quantum drug-design programme, which we wrote about on 19 August as an argument about verification, runs across a Yonsei-hosted IBM quantum environment, cooperation with Quantinuum, NVIDIA and D-Wave, and a domestic synthesis partner. The field treats that spread as ordinary. Read against this month’s documents, it is also a map of exposure.
Practical takeaway. Export-control law has treated people as a transfer channel for a long time, and neither new measure changes the fundamental-research exclusion that keeps most university work outside it. What they add is narrower than blanket personnel control and lands in two new places: the institutional partnership and the funding attached to it, and an individual’s ability to leave. Co-authorship by itself is not the trigger on either end. A collaboration agreement signed in 2024 or 2025 now carries affiliation and mobility consequences written after it was signed.
What this feels like from inside a career
Instruments of this kind are usually discussed in the language of leverage and deterrence. Their texture is smaller and more human. A researcher with a joint appointment finds that one line on a CV, added years ago at the encouragement of both employers, is now the line the audit is looking for. A doctoral student whose supervisor’s former laboratory appears on a list learns that the mitigation plan concerns them. Somebody books a conference flight without being able to find out whether a decision has been recorded against their name; the Chinese provisions allow notification to be withheld where national security or a criminal investigation is involved.
A shipment can be recalled and a licence can be amended. An affiliation cannot be unwritten. That is the durable difference between a control aimed at goods and one aimed at institutional ties: it attaches to a record that only accumulates. Neither government is acting frivolously or secretly. Washington is protecting research it paid for; Beijing is asserting a broader claim over technology its citizens carry, whoever funded it. The cost lands on individuals who made ordinary career decisions under the previous rules, and it lands unevenly, on the researchers with the most international training, who tend also to be the best.
The institutional cost is quieter. Universities were designed to be porous, and that porosity is where their productivity comes from. A forensic review of foreign entanglements inside a fortnight, with research funding as the stake, asks an institution to work against its own instinct at speed. Some partnerships that would survive careful scrutiny will be cut anyway.
How Quentir Reads It
The thing to hold onto is the change of object, not the temperature of the rhetoric. For most of the modern history of technology control, the question a compliance function asked was what is leaving. The instruments above add a second one: which affiliations are on our books, and which now carry a consequence. That is answered from different records, on different timescales, and most compliance functions were not built to answer it — a narrower change than research-security rhetoric usually suggests, and a more durable one.
It also changes what a research collaboration is worth. In our reading of what DARPA was actually buying we argued that funders are increasingly buying reproducibility where they used to buy novelty; the same week we looked at the state taking equity positions in quantum firms. The American measure belongs beside those two as a third way of shaping a research base through its funding, and much the cheaper one: no appropriation, no equity stake, only a list and a deadline. The Chinese measure is not in that family at all, which is why the pair repays reading together — the same pressure, applied through money on one side and through movement on the other.
Every post we publish on this becomes part of one running record, and the connections between them are the point: the All-access membership keeps that archive in one place, so a July instrument can be read against an August order without reassembling a year of posts.
The near-term date to watch is 15 September — not for any drama on the day, but because the Chinese provision publishes no detailed application criteria beyond its broad trigger and sets no fixed duration. Its real scope will be set by whoever is stopped first. Whether that person turns out to be a semiconductor engineer, a battery chemist or somebody from a quantum group will say more about the perimeter than the nineteen articles do.
Sources: State Council of the People’s Republic of China, “China unveils regulation on exit, entry administration”, 31 July 2026, for the publication date and the 15 September 2026 commencement. DLA Piper, “China’s new exit and entry regulation: what multinational companies need to know”, 4 August 2026, for the nineteen-article structure, the export-control exit-ban trigger wording, and the absence of a fixed time limit for that category where others run six months to three years. South China Morning Post, “Beijing to impose exit bans for export control, tech transfer breaches”, 31 July 2026, for the enforcing departments. The full Chinese text of the Provisions was not retrieved for this post; the provision is described here as the official release and these two readings of it describe it. Jon Harper, “Pentagon probes 30 U.S. academic institutions for potential ties to adversaries”, DefenseScoop, 17 August 2026, for the issuing office, the Section 1286 authority, the thirty institutions, the Confucius Institute scope, the 31 August 2026 deadline, the “termination of problematic partnerships” mitigation language, the funding consequence and the approximate size of the fiscal 2025 list. The Department of War’s own public release of the same date is the primary announcement of the audits; its full text was not retrieved for this post, so the operative details above are given as DefenseScoop reports them. The individual notifications are not public and do not enumerate technology fields; the quantum connection in this post rests on the strategy document, not on the audit order. Office of Science and Technology Policy, National Security Science and Technology Strategy, August 2026, for the fourteen critical and emerging technology areas including quantum information technologies and for the FBI-led Quantum Information Science and Technology Counterintelligence Protection Team; see also the Federal Bureau of Investigation, “Protecting Quantum Science and Technology”. The 17 August release date of the strategy is taken from Nextgov/FCW, “White House S&T strategy calls for new approaches to safeguard US research”, 17 August 2026, which reports the document as publicly released that Monday. Bureau of Industry and Security, “What is a Deemed Export?”, for the deemed-export rule and its scope, including the fundamental research, published information and immigration-status exclusions described in the opening. Public sources checked 20 August 2026.
Published intelligence, built to inform your own decisions. Published: August 20, 2026.