What Open Means for Science in a Closing World

Governments are limiting who scientists can work with, often without weighing what those limits cost.

November 17, 2026
Sanders, Barry - What Open Means for Science
Wafer production in a clean room at the quantum computing start-up Peak Quantum in Munich, Germany. (Angelika Warmuth/REUTERS)

Scientists hold dear the notion of openness, a notion shared by the agencies funding science. Recently, though, restrictions on open science have begun to appear, particularly because of security considerations. These restrictions became particularly pronounced with the July 2026 announcement that the US National Science Foundation (NSF) would implement a policy commencing in the fiscal year 2027 on collaboration with restricted entities and their employees. This announcement is part of a global seismic shift that is raising boundaries for global science and causing scientists to rethink how to conduct and share scientific research.

The case for openness is one that various agencies have long made. In the United States, for example, the National Institutes of Health articulated in 2023 that “sharing data accelerates biomedical research discovery,” with rigour and reproducibility built on the edifice of sharing. The intergovernmental European Organization for Nuclear Research (CERN) holds the core value of making “research open and accessible for everyone.” The United Nations Educational, Scientific and Cultural Organization recommends open science in a broad sense that includes publications, data, software, infrastructure and engagement. Through these open practices, science brings benefits (and harms) to all societies, while also holding the field accountable to other scientists and to the world.

My scientific research focuses on quantum computing. Quantum computing might deliver revolutionary change by transforming certain intractable (hard-to-solve) computational problems into efficiently solvable problems, which is why it is now regarded as a sensitive technology. Canada’s Policy on Sensitive Technology Research and Affiliations of Concern (STRAC) restricts openness on quantum computing and other sensitive technologies, including artificial intelligence (AI) and genetic engineering. Similar to the NSF’s new policy, STRAC restricts Canadian researchers in sensitive technology areas from having affiliations with, or cash or in-kind funding from, certain named research organizations.

By contrast, Canada’s National Quantum Strategy Consultations emphasize the vital importance of international collaboration, admission of foreign talent into Canadian research and development endeavours, and sharing of big infrastructure in order for Canada to succeed in being a major force in quantum technologies. Consequently, the costs of restrictions to scientific openness are not simple or straightforward; they must be weighed and evaluated, partly because the collaboration and sharing of talent and infrastructure are key to the scientific process of verifying and validating, or even refuting, claims.

These costs to scientific openness are augmented when restrictions over funding and affiliations are accompanied by export controls, which limit the permitted flow of equipment and technology to other countries. In July 2024, Canada implemented export controls for quantum computing, with exceptions such as for the United States. Allied countries announced similar measures around the same time, suggesting a coordinated effort to control advanced quantum computing. The irony is evident here, as the results of using quantum computing could be reported in open publications but the tools needed to test and reproduce those results are not made widely available.

The Limits of Openness

Although restrictions are growing, science has not always been fully open. For example, CERN’s data policy permits timed releases of particle collision data and explains that raw data sets are not necessarily meaningful outside the collaborations that produce and interpret such data. CERN makes reconstructed data and associated software and documentation available, and this form of data sharing is arguably more valuable to independent researchers than impenetrable raw data.

In practice, however, science is highly interrelated with technology, and one can be hard to distinguish from the other as scientific interests embody engineering, software and experimental techniques. At the same time, keeping science open can involve protecting and hiding advances in technology and engineering. A scientific result can depend on reporting achievements using a novel instrument, but the design or right to use that instrument can be concealed. Although sharing designs could accelerate development and diminish obstacles, a balancing act is needed as this openness risks compromising intellectual property (IP) or revealing sensitive capabilities. Even beyond concealing an instrument’s design, allocating access to the instrument might need to be done in a seemingly inequitable way to account for various needs and restrictions.

Horizon Europe, the European Union’s major research funding program, mandates the open science values of open access to publications and data. That said, the agency articulates the principle of being “as open as possible, as closed as necessary” regarding data access. The starting point is being open, with closedness imposed for legitimate reasons such as privacy, IP and security. Such restrictions are expected to be articulated in a data management plan, but the devil is in the details, and it is difficult to define what needs to be protected, what evidence is required to justify such protections and how to weigh the associated costs. For example, patient records can be kept confidential, but the way that the data is used and analyzed can be openly published.

Openness is an issue involving not just disclosure, but also governance frameworks. Peer review is a long-standing method for evaluating scientific merit, whereas funding organizations, institutions and security agencies are positioned to decide which areas, entities and researchers to support. Decisions about who to support, as well as how much and under what conditions, have wide ramifications not just on science but also on a nation’s economy and security. Such decisions need to be approached with eyes wide open, determining the benefits and harms of openness and its restrictions. Another important consideration is being transparent about any restrictions imposed, including due process and time constraints (and how to contest the latter).

In response, my colleagues and I are developing a framework for assessing and guiding openness at the Centre for International Governance Innovation. Our framework examines openness by first identifying the object under consideration — for example, a design for a nuclear reactor, a genomic sequence or a release of open weights for AI — and then describing it in terms of dimensions such as power, access, practice and benefits. This description of the object comes before articulating principles such as transparency and openness and values such as equity and accountability to ensure that the facts are in plain view before deciding the degree of openness. In this assessment, it is also important to decide what constitutes evidence and who makes that decision.

The COVID-19 pandemic and subsequent rancorous debate over vaccines offer a salient example, prompting questions such as what constitutes evidence of benefits and harms and who makes these decisions, how claims can be contested, and so on. The highly respected British Medical Journal wrote in January 2022 that “Data should be fully and immediately available for public scrutiny” and that “transparent decision-making is essential.” Besides sound scientific reasoning, sound openness principles are arguably just as important in managing innovation.

Scientists and policy makers need to collaborate on openness — and on closedness — in science and technology. Openness can advance discovery and promote public benefit, but it also brings economic and security risks. Restrictions can protect legitimate interests while also impeding scrutiny and equitable access to knowledge. Getting that balance right is how science can remain open in a closing world.

AID Statement: Artificial Intelligence Tool: ChatGPT by OpenAI, used in October 2026; Information Collection; Conceptualization.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

About the Author

Barry C Sanders is a CIGI senior fellow and a leading figure in quantum science and technology, with research spanning quantum sensing, communication, computing and optics.