Skip to content
INTERNET INFRASTRUCTURE & NETWORKS

Preparing for the Post-Quantum Era: How Cloudflare and Partners Are Revolutionizing Anonymous Credentials for the Web

As the technology industry looks ahead to the arrival of cryptographically relevant quantum computers, the primary focus for cybersecurity experts has largely centered on the foundational building blocks of secure internet communication. Protocols like TLS, which rely on quantum-vulnerable cryptography, represent the most urgent priority for transition. However, internet infrastructure depends on classical cryptography for far more than just basic encrypted transport. Protecting modern web applications from abuse, managing automated traffic, and preserving user privacy all rely heavily on cryptographic tools that will become completely insecure once "Q-day" arrives.

Among the systems facing this impending cryptographic obsolescence is Privacy Pass, a protocol widely deployed by site operators such as Cloudflare to protect web applications from malicious activity and automated abuse. A central function of Privacy Pass is rate-limiting individual clients to prevent any single source from consuming an unsustainable share of a site’s resources. Traditionally, this type of traffic management forces users to endure friction or sacrifice their privacy through the use of intrusive CAPTCHAs, traffic challenges, account logins, or browser fingerprinting. Privacy Pass was specifically engineered to bypass this false trade-off between user privacy and operational security.

The underlying mechanism relies on tokens issued to a client, which can later be redeemed to gain seamless access to a site. Crucially, these tokens are cryptographically unlinkable to the specific client they were originally issued to. This ensures that legitimate human visitors can navigate the web safely without abandoning their right to privacy. With a decade of deployment experience supporting applications like privacy proxies, including iCloud Private Relay, and traffic relays like Privacy Gateway, developers and industry collaborators are now working to bring the next generation of this technology to the broader web.

This collaborative effort targets two main objectives: building more sophisticated tools to manage agentic traffic—automated systems and AI agents that can degrade web performance for human visitors—and fostering a more open ecosystem that allows a wider variety of client types to participate. Achieving these ambitious goals requires developing brand-new cryptographic capabilities that extend far beyond the limits of current Privacy Pass implementations.

Fortunately, the current trajectory of post-quantum research offers a glimmer of hope, putting the industry in a slightly better position than many might initially assume. Privacy Pass currently relies on classical cryptographic primitives, such as RSA and elliptic curves. This means that a quantum-enabled attacker will eventually be able to forge tokens at will, completely bypassing site moderation and rate-limiting controls. On the other hand, Privacy Pass possesses a vital property: it is unconditionally private. No attacker, quantum or otherwise, can link the redemption of a token back to its original issuance. Furthermore, the advanced capabilities being developed for the next generation of private authentication technology are being designed to uphold this exact same guarantee, ensuring that user interactions remain entirely private long after the arrival of quantum computing.

This inherent separation of concerns provides a modest cushion of time, but viable post-quantum alternatives must be established soon. While researchers have made significant strides, the primary challenge mirror the broader post-quantum transition: the sheer physical size of post-quantum cryptographic primitives. For instance, an ML-DSA-44 signature measures a massive 2,420 bytes, whereas a traditional ECDSA-P256 signature is a mere 64 bytes. Because seamless drop-in replacements are unlikely to emerge anytime soon, engineers must accept primitives with higher operational costs and fundamentally re-engineer their systems to bear that weight.

What We Need to Build: A Brief Look at MOLE

To understand how the web might achieve post-quantum privacy, it helps to examine an emerging technology framework currently known as MOLE, an acronym for "Moderation Of unLinkable Endorsements." At the core of traditional Privacy Pass deployments is a blind signature scheme. This cryptographic primitive allows a client to obtain a signature under an issuer’s secret key without ever revealing the underlying message to the issuer. The client can subsequently present the message and the blind signature later without establishing any link to the initial issuance phase.

In a standard Privacy Pass token, the asset consists of a message combined with a blindly issued signature. To verify the token, the receiving server must evaluate the issuer’s public key, which inherently exposes the identity of the issuer. While this design protects the identity of the client, it fails to protect the identity of the issuer. For certain use cases, this trade-off is acceptable because the issuer’s identity signals what type of client is connecting to the network edge, helping businesses verify authorization for specific enterprise products or virtual private networks.

However, for many common web applications, revealing the issuer’s identity provides far more information than necessary. Generally, a site operator only needs to know that a token was issued by some trusted entity, rather than a specific one. MOLE takes this concept a step further by entirely hiding the issuer’s identity, preventing web operators from treating incoming traffic differently based on who issued the credentials. This enforces equal treatment across different client types, lowering barriers to entry and expanding ecosystem participation.

This issuer-hiding property is achieved through a newly conceptualized cryptographic object called an endorsement. An endorsement is issued by an anchor and ultimately redeemed by a moderator. Before redemption, the client and the moderator agree on an anchor set, which comprises the public keys of all anchors trusted by the moderator. During the redemption process, the client proves that it received an endorsement from at least one of the anchors in the anchor set, while successfully concealing which specific anchor performed the issuance. Once accepted, the moderator issues the client a MOLE credential used to govern subsequent interactions with the website.

Because these credentials are intended for use by a specific moderator, they do not need to be anchor-hiding or publicly verifiable. This flexibility allows engineers to instantiate MOLE credentials using a wide array of underlying primitives, ranging from traditional blind signatures to anonymous credit tokens and advanced rate-limited credentials.

A Feast of Cryptography

In cryptographic protocol design, adaptability and efficiency are frequently at odds. Tailoring a construction for a hyper-specific use case yields high efficiency, but it diminishes the protocol’s adaptability when circumstances change. For a system like MOLE, adaptability is non-negotiable. Web traffic patterns and attacker behaviors evolve continuously, requiring a versatile suite of building blocks that can be composed into diverse protocols to meet shifting security

demands.

Most modern post-quantum cryptographic schemes derive from larger mathematical families of constructions that have been heavily optimized for single applications. ML-DSA serves as a prime example; its underlying mathematical framework is flexible, but its specific parameters were carefully tuned to produce an exceptionally compact and efficient digital signature scheme. Those exact parameters, however, would not necessarily remain secure if applied to alternative use cases.

The central challenge in designing post-quantum anonymous credentials lies in successfully decoupling parameter selection from protocol design. Classical credential systems afford engineers the luxury of ignoring the underlying mathematical parameters of their elliptic curves, allowing non-experts to build secure applications. Designing efficient post-quantum credentials, by contrast, has historically demanded an extraordinary depth of specialized cryptographic expertise.

The post-quantum future of web privacy

VOLE-in-the-Head

A major cryptographic breakthrough introduced a potential sea change for anonymous credentials by bridging advanced zero-knowledge proofs with multivariate mathematics. The conceptual roots of this development trace back to the VOLE-in-the-Head, or VOLEitH, zero-knowledge proof system introduced by academic researchers at Crypto 2023. A zero-knowledge proof system allows a prover to demonstrate knowledge of a hidden witness that satisfies a public statement without revealing any information about the witness itself.

For example, a zero-knowledge proof can confirm knowledge of a hidden AES encryption key linked to a known public output, effectively transforming a symmetric cipher into a digital signature scheme whose security relies directly on AES. Compared to many alternative proof systems, VOLEitH is relatively straightforward to understand and implement securely. The acronym stands for "Vector Oblivious Linear Evaluation," an interactive technique that correlates random values between two parties to act as authentication tags, allowing verifiers to confirm that computations on hidden data were executed correctly.

VOLEitH operates by having the prover locally simulate the values that two parties would normally generate interactively—hence the moniker "in the head"—before committing to those values and applying the Fiat-Shamir transformation to generate a non-interactive zero-knowledge proof. While VOLEitH proofs are asymptotically linear in the size of the witness and lack the succinctness found in other proof systems, the framework is concretely efficient for statements expressed as equations over small binary fields.

One prominent application is the FAEST digital signature candidate currently evaluated in the ongoing NIST signature standardization competition, which handles symmetric cryptographic relations efficiently. Another application is the KuMQuat signature scheme, where the signer proves an instance of the Multivariate Quadratic problem over small fields like single-byte binary fields. The MQ problem involves solving systems of random quadratic equations across multiple variables, a mathematical hurdle believed to remain secure even against quantum adversaries. Because multivariate maps rely on quadratic equations, proving solutions using VOLEitH is comparatively inexpensive, yielding compact signature sizes.

Multivariate Signatures

Alternative methods for constructing signatures from multivariate quadratic maps involve embedding specialized mathematical structures directly into the map. The Unbalanced Oil and Vinegar signature scheme uses an MQ map as its public key, where a signature consists of a vector satisfying a target vector derived from hashing and salting the signed message. The secret key contains a trapdoor that allows the signer to partition variables into "oil" and "vinegar" variables, simplifying the quadratic equations into a solvable linear system.

While UOV yields remarkably small signatures, it suffers from exceptionally large public keys. Conversely, alternative multivariate candidates like MAYO achieve much smaller public keys at the cost of moderately larger signatures. Crucially, the arithmetic complexity of evaluating a UOV or MAYO signature matches the complexity of evaluating a general MQ instance, making them natural candidates for integration with VOLEitH frameworks.

Blind Signatures from VOLE-in-the-Head and UOV

Recent academic literature introduced PoMFRIT, a novel blind signature construction combining VOLE-in-the-Head with multivariate signatures to achieve highly competitive performance metrics for post-quantum parameters. By swapping specific multivariate components, developers can efficiently tune signature sizes down to compact footprints, trading slightly larger public keys for massive efficiency gains in environments where public keys can be cached and reused across numerous sessions.

The architectural paradigm relies on clients placing messages inside cryptographic commitments—essentially sealed digital envelopes—allowing signers to sign the commitment directly. The client then generates a zero-knowledge proof demonstrating knowledge of the valid signature and the opening randomness. Because the client presents a commitment during issuance and an entirely separate zero-knowledge proof during presentation, unlinkability is guaranteed even if the issuer and verifier collude completely.

Optimizations and Security Trade-Offs

Achieving practical performance in post-quantum cryptography often requires balancing conservative security postures with clever engineering optimizations. Researchers have explored strategies such as reducing the number of cryptographic hash permutations required within the zero-knowledge proofs and shrinking internal state sizes to minimize algebraic constraints. By optimizing Keccak permutations and adopting leaner round structures, developers can significantly slash proof sizes without meaningfully compromising the underlying security margins.

At the same time, multivariate cryptography remains a relatively young discipline compared to traditional number-theoretic constructions. Although UOV and MAYO have demonstrated resilience against recent cryptanalytic attacks, ongoing study is vital before deploying these systems at scale. Furthermore, cryptographic engineers must evaluate whether optimizations like pre-image decoupling introduce subtle risks that could affect forgery resistance, while keeping in mind that any potential failure in these mechanisms would impact signature integrity rather than the absolute privacy guarantees protecting individual clients.

Is MOLE Ready for Post-Quantum?

Because classical privacy mechanisms remain operational today, the cryptographic community retains a valuable window of time to mature these alternatives before quantum computers pose an immediate operational threat. Multivariate cryptography is rapidly proving to be a versatile foundation for constructing a wide array of anonymous tokens and credentials. Ongoing development efforts continue to refine candidate protocols for MOLE endorsements and credentials, steadily optimizing performance metrics and communication overhead.

While post-quantum primitives naturally introduce larger data footprints than their classical predecessors, the computational runtime of these advanced protocols remains exceptionally fast, requiring just a few milliseconds per operation in practical implementations. As the transition moves forward, the primary engineering challenge will shift from fundamental cryptographic design to practical protocol optimization, such as amortizing redemption costs across multiple requests.

With foundational research advancing rapidly and implementation frameworks already taking shape, the transition to post-quantum anonymous credentials for the web is well underway. As ongoing cryptanalytic study stabilizes the underlying parameter sets, these robust multivariate schemes are positioned to secure user privacy across an expanding ecosystem of internet applications long after the arrival of quantum computing.