TL;DR
A paper in the Journal of Quantum Information Science argues that AI could contribute both to advances relevant to quantum attacks on encryption and to defensive work such as finding vulnerable cryptography. It presents a strategic framework, not new experimental results, and says the timing and scale of future quantum threats remain uncertain.
A new analysis published in the Journal of Quantum Information Science argues that artificial intelligence could help advance quantum capabilities that threaten some widely used encryption while also helping organizations identify vulnerable systems and plan replacements. The paper presents a cybersecurity planning framework, not new experimental findings, and says organizations should prepare without relying on a fixed date for a capable quantum computer.
The analysis, by Troy Coienth Troublefield of Capitol Technology University, focuses on public-key cryptography such as RSA and elliptic-curve systems. These methods support secure connections and digital identity checks. A sufficiently capable quantum computer running Shor’s algorithm could undermine the mathematical problems on which those systems rely. The paper says current quantum processors remain too small and error-prone to attack widely used cryptography.
AI could contribute to quantum development by helping researchers improve quantum error-correction decoders, software that interprets measurements used to identify errors in calculations. The paper also discusses changing estimates of the resources needed to break RSA-2048. A 2025 estimate by Google researcher Craig Gidney put the requirement below one million physical qubits, down from an earlier estimate of about 20 million. These are theoretical estimates under specified assumptions, not demonstrations of an attack or a simple qubit threshold. Troublefield attributes the revisions to a range of developments, including more efficient algorithms and changes to error-correction methods—not AI alone.
The paper identifies defensive uses that may be more immediate: automated tools could help organizations locate cryptography across applications, code libraries, certificates, connected devices and supplier services. That inventory can help security teams prioritize replacements and track changes as systems are updated. Some tools search for known cryptographic identifiers; others use machine learning to infer cryptographic use from code or network patterns. The analysis says independent public evaluations of commercial tools’ accuracy, coverage and false alarms remain limited.
Why Long-Lived Data Needs Attention
The main planning concern is not only when a quantum computer capable of breaking current public-key systems might arrive, but how long sensitive information must remain protected. Under the “harvest now, decrypt later” scenario, an attacker collects encrypted information today and attempts to read it in the future. Data with value lasting decades—including medical records, intellectual property and classified material—could remain exposed if it is intercepted before organizations complete a cryptographic migration.
The analysis says this risk can justify preparation even while the development timeline is uncertain. Replacing cryptography across large organizations takes time because security teams must find where it is used, understand dependencies and coordinate with vendors. The paper also calls for including AI training data, model files and information handled by autonomous software in migration planning, rather than treating AI systems as separate from the wider security inventory.
AI therefore appears in the paper as both a possible contributor to the threat and a potential aid to defense. That does not establish that AI has already enabled a practical quantum attack. It does underscore the need to distinguish projected capability from current evidence while organizations assess which information would be costly if exposed later.
quantum cryptography security tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Quantum Estimates and Migration Planning
The cryptographic risk described in the paper depends on a future machine able to perform reliable, large-scale calculations. Conventional computers cannot practically solve the relevant mathematical problems at the sizes used by RSA and elliptic-curve cryptography, while a sufficiently capable quantum computer could use Shor’s algorithm against them. The paper says present-day processors do not have the scale and reliability for that task.
Resource estimates are not proof that such a machine is close. Gidney’s 2025 estimate for RSA-2048 was lower than an earlier estimate of about 20 million physical qubits, but the figures depend on assumptions about algorithms, hardware and error correction. The analysis cautions against attributing the change to AI alone. It instead uses existing research, technical reporting, government standards and security assessments to argue for preparation as a continuing process.
Government agencies have warned about the possibility of collecting encrypted data for future decryption. The paper says publicly available evidence does not establish who is collecting such data or how much. Its proposed response is to identify vulnerable cryptography, prioritize information by how long it needs protection and move toward quantum-resistant methods.
“The paper calls for treating AI and quantum computing as a single cybersecurity challenge.”
— Troy Coienth Troublefield, Capitol Technology University
AI-powered cryptography vulnerability scanner
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Timing, Collection and Tool Accuracy
The paper does not establish when a quantum computer capable of breaking widely used public-key cryptography will be available. Its discussion of AI-assisted error correction and lower resource estimates describes potential contributions and theoretical requirements, not a demonstrated path to an operational attack. It also does not show that AI alone is driving progress toward such a machine.
Public evidence cited in the analysis does not establish whether attackers are currently collecting encrypted data at scale or identify the amount collected. The practical performance of commercial cryptographic discovery tools is another open issue: the paper says public, independent assessments of their accuracy, coverage and false-alarm rates are limited. The available source material also does not specify a universal migration timetable or quantify the cost and duration of organizational transitions.
As an affiliate, we earn on qualifying purchases.
Inventory Before Cryptography Changes
The analysis recommends that organizations begin by finding where cryptography is used, then rank systems according to the sensitivity and required protection life of the information they handle. Security teams can use that inventory to plan replacement with quantum-resistant protection, coordinate changes with suppliers and revisit priorities as systems evolve. The paper also recommends including AI assets such as training data and model files in that work.
The next practical test is whether automated discovery tools can provide reliable coverage across complex environments. Organizations will need to validate results and account for missed systems and false alarms, particularly given the limited public evaluation described by the paper. Further research and clearer evidence about quantum hardware, tool performance and data collection could refine planning, but the study’s central recommendation is to treat preparation as ongoing rather than wait for a settled forecast.
cryptography inventory management tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
Does the study report a successful quantum attack on encryption?
No. It is a strategic analysis, not a report of a new attack or experimental result. The paper says current quantum processors are not capable of attacking widely used cryptography.
How could AI contribute to quantum threats?
The paper suggests AI may help researchers improve quantum error-correction decoders, which could support more reliable quantum calculations. It does not claim AI alone will make a cryptographic attack possible.
What does “harvest now, decrypt later” mean?
It describes collecting encrypted information now and keeping it in the hope that a future quantum computer could decrypt it. The analysis says public evidence does not establish who is collecting such data or how much.
What can organizations do now?
The paper recommends locating cryptography across systems, prioritizing information that needs long-term protection and planning a move to quantum-resistant methods. It also says AI-related data and model files should be part of that inventory.
Are commercial tools for finding vulnerable encryption proven accurate?
The analysis says independent public evaluations of commercial tools’ accuracy, coverage and false alarms remain limited. Automated discovery can support an inventory, but the paper does not establish that such tools are uniformly reliable.
Source: rss
