Over the recent years, we have seen the consequences of unregulated AI growth play out in front of our eyes. Eager to avoid repeating the same mistakes, policymakers have began the push towards quantum computer (QC) regulation. Many countries have brought about initial QC export controls, with others deemed to follow suit.

While welcome, this move likely does not go far enough to absolutely avert all the possible dangers of QCs. In this post, I list additional policy proposals for bold, forward-looking regulators willing to ensure a safe, equitable quantum future for all.

Import controls. The idea is simple - under the proposed policy, any QC advanced enough to be covered under an export control will also be protected from foreign interference through import controls. Specifically, every component of foreign origin used in the construction of the QC must be registered and approved for import. The approval will be granted swiftly to any part with a clean provenance. However, should the supply chain involve China in any way, further scrutiny will be required, and the purchasing company will be asked to present evidence that no suitable alternatives could be identified.

Certified quantum engineers. Do you know why - with rare exceptions - trains don’t fall off tracks and cars don’t disintegrate mid-journey? It’s because only engineers certified by appropriate professional bodies can oversee their design, construction and deployment. By contrast, since anyone can call themselves a quantum engineer, is it any surprise that qubits decohere mid-computation? To avoid further chaos, we propose that only certified quantum engineers be allowed to develop quantum computers. A government-approved certification body will be set up to ensure quantum engineers undergo regular training on the latest developments in the field, and are up to date with the growing body of regulations. Stringent controls will be placed at the point of QC sale to ensure compliance.

One temperature for all qubits. In today’s fragmented QC ecosystem, different qubits operate at different temperatures. To make matters worse, even cryogenic qubits operate at vastly different base temperatures, ranging from as low as 100 uK to as high as 77 K. This market segmentation makes business and regulation very challenging. We propose that a temperature standardisation panel is formed by representatives of the QC industry, academia, and policy space. Following its recommendation, all QC platforms will be required to adopt a standard operating temperature. This rule will only apply to qubits – other ancillary equipment will be still allowed to operate in any temperature. Once codified as an ISO standard, this policy will reduce requirements dilution, enabling rapid progress in the space of cryocooling.

Universal noise injection. Governments worldwide should sign an agreement to inject broadband noise in the 10 MHz – 10 GHz band into the power grid. As a result, all grid-connected quantum computers will experience additional decoherence, preventing them from reaching fault tolerance. Third-party inspectors will be sent unannounced to verify national grid compliance. This measure will buy extra time for society to have a nuanced conversation about the pros and cons of QC. When the time is right, the noise magnitude can be gradually reduced, following a recommended schedule agreed by an international panel of experts.

It would be interesting to consider alternative implementation methods that can achieve the same goal. For example, noise could be also injected into the 4G/5G network radio waves, and pollutants could be sprayed into cleanroom HVAC systems preventing high-quality qubit production. If readers have other ideas, please post them in comments!

Nuclear qubit restrictions. Nuclear states can store quantum information with long lifetimes (by QC standards). This raises questions about their dual-use capabilities and impact on health. Until better studies are available, we propose that all matter-based QCs should operate on electronic states only. Qubit encoding into nuclear states will be restricted without explicit permission, and placed under the jurisdiction of the nuclear energy authority. The proposal extends to any system with significant electron-nucleus coupling, and restricts quantum information storage in nuclear states, hybrid electron-nucleus states, as well as the use of dynamical decoupling to protect the electron from nuclear interactions. It is recommended that nuclear spins are eliminated from the QC system altogether, as long as that’s achieved without the use of isotropic material purification, another restricted technique.

Quantum circuit registration. Any day now someone somewhere may run the first classically-intractable quantum computation of geopolitical importance. To avoid the dangers of this free-for-all, governments should mandate that all quantum compilers incorporate a “registration pass”. During the “registration pass”, the circuit is sent to an appropriate government agency to verify its societal impact before approving or rejecting its execution.

Several measures should be undertaken at the same time to avoid excessive slowdown of scientific progress. First, a new department of the civil service should be appointed to manage the workload. Second, “trusted circuit providers” could be identified in due course. For these “trusted providers”, universal circuit verification could be replaced with circuit spot checks. Third, a system could be set up to automatically approve circuits which are only small variations on previously approved circuits (e.g. different parameter sets for a VQE routine). Assuming those measures are put in place, I am confident that the registration pass would take no more than 2-4 weeks for the average user.

Sustainability first. Most QC hardware today involves aluminium and steel, whose manufacture is a major contributor to climate change. Furthermore, ancillary control and laboratory equipment frequently use plastics full of PFAS. Going forward, QC hardware must transition to more ecological materials, such as bamboo and beeswax. All plastics must be biodegradable and recycled.

Climate-neutral QCs. QC presents itself as a solution to climate change, both due to its potential to revolutionise biochemistry, as well as its inherent energy efficiency. It’s therefore fair to require that QCs themselves become climate-neutral from the get-go. Under this proposal, all QCs should be operated exclusively using renewable energy. Furthermore, all cryocooling systems should be turned off whenever no quantum gates are applied, as well as over the summer, when the demand for residential air conditioning is at peak capacity.

Autonomous QC? No thank you. Most existing proposals for quantum error correction (QEC) involve a classical chip decoding errors and applying correction operations. However, a recent stream of research is looking to implement QEC autonomously, without human oversight. We believe that granting novel, experimental computing systems such a degree of autonomy is misguided, and should be outlawed as a rule. The same reasoning applies to QCs controlled by AI agents, obviously.

Cat qubits research should require animal studies approval. Plus, cat are vicious, so why play with fire?

No counter-rotating terms. Traditionally, QC control is derived assuming the rotating-wave approximation. The resulting Hamiltonians are well-behaved and analytically simulatable. More recently, a wave of papers proposes to employ the fast-rotating terms - and even the counter-rotating terms - to create more powerful qubit control schemes. Such methods cannot be understood analytically, thus their effect is a priori unpredictable. We believe that QCs should not rely on fast-rotating terms, and public policy should cement that as law.

Private patents. Industrial QC players require robust patent enforcement to incentivise R&D efforts. However, the QC patents pose a national security threat, as they disclose in the open the secrets of manufacturing and operating quantum computers. To resolve these conflicting requirements, we propose that QC patents be granted as usual, but remain private for an indefinite period. Any patent litigation must be conducted behind closed doors, with strict access restriction to the patent documents.

Your idea? Do you have any killer ideas for QC regulation? Let me know in the comments!