Cloudflare Launches OHTTP Gateway: New Infrastructure for Privacy Protection

Cloudflare productizes OHTTP gateway, letting developers decouple user identity from request content without building their own infrastructure.
Oblivious HTTP (OHTTP) is an IETF-standardized privacy protocol that inserts independent relay and gateway intermediaries between client and server, ensuring no single entity can link a user's IP to their request content. Cloudflare's new managed OHTTP gateway significantly lowers the integration barrier for developers. The privacy guarantee depends on the relay and gateway not colluding, and the protocol is best suited for one-way scenarios like app telemetry and password breach detection. Limitations include added latency and the fact that request content remains visible to the gateway itself.
What Is OHTTP
Oblivious HTTP (OHTTP) is a network protocol designed to decouple user identity from request content. In a traditional HTTP request, the server receiving the request can see both the content of the request and the client's IP address — meaning it can link "who" with "what they did" to build user profiles. OHTTP's core idea is to insert a "relay" and a "gateway" in the middle, so that no single entity can simultaneously possess both types of information.
Specifically, the client encrypts the request content first, then forwards it to the gateway through a relay. The relay can see the client's IP address but cannot read the encrypted request content; the gateway can decrypt and process the request content but only sees the relay's IP, with no way to identify the original client. As long as the relay and gateway are operated by different, non-colluding organizations, the user's identity and request content are effectively isolated.
Cloudflare's newly announced OHTTP gateway service productizes the "gateway" role in this mechanism, lowering the barrier for applications to integrate privacy-preserving capabilities.
OHTTP's encryption mechanism is based on HPKE (Hybrid Public Key Encryption). The client uses the gateway's public key to asymmetrically encrypt and encapsulate the request, ensuring the relay cannot read the request body during forwarding. The entire protocol was formally standardized by the IETF in RFC 9458, with one of its design goals being the minimization of metadata leakage — even the relay can only observe that "some IP sent a request to some gateway," with no access to any request content. This "blind forwarding" characteristic fundamentally distinguishes OHTTP from traditional VPNs or HTTPS proxies: a VPN provider can simultaneously see both user identity and traffic content, whereas OHTTP architecturally prevents single-point awareness at the protocol level.

Why This Matters
OHTTP is not an entirely new concept — it was standardized by the IETF, and Apple's iCloud Private Relay and certain telemetry use cases have already adopted it. However, deploying OHTTP has long required developers to build and maintain their own gateways, creating a high barrier to entry. By offering it as a managed service, Cloudflare makes this infrastructure directly accessible to far more developers without requiring them to build from scratch.
From an industry perspective, the pressure around privacy compliance is growing. App telemetry, crash reporting, and usage analytics all face the tension of wanting to collect data without invading user privacy. OHTTP offers a middle path: applications can still obtain aggregate-level statistics, but cannot trace individual data points back to a specific user. This is particularly attractive to vendors that need to collect sensitive telemetry.
On Hacker News, the announcement received 183 upvotes and nearly 90 comments, reflecting sustained developer community interest in privacy infrastructure.
The Trust Model Between Relay and Gateway
The key to OHTTP's privacy guarantee lies in "separation of duties" — the relay and gateway must be operated by different entities that do not collude. This is a point repeatedly raised in discussions: if the same company operates both the relay and the gateway, it could theoretically combine information from both sides, rendering the privacy promise meaningless.
Therefore, a sensible deployment model when using Cloudflare's gateway service is to have an independent third party operate the relay. This architecture resembles Apple Private Relay's two-hop design, where Apple handles one hop and a third-party CDN handles the other, creating a system of checks and balances. For developers adopting Cloudflare's gateway, choosing a relay provider with no business ties to Cloudflare is a prerequisite for the privacy model to hold.
It's important to recognize that OHTTP provides privacy protection under the premise of non-collusion — not absolute anonymity in the cryptographic sense. Its security is built on distributed trust among operators, making it a privacy solution that combines engineering with governance.
The "non-collusion assumption" has a long history of discussion in cryptography and privacy engineering. It is essentially a boundary condition of a threat model: when two operators have a competitive relationship, are subject to different jurisdictions, or face different regulatory constraints, the likelihood of collusion is relatively low. Apple's choice to partner with CDNs like Akamai and Cloudflare for the second hop of iCloud Private Relay is precisely an attempt to reinforce this guarantee through commercial logic and geographic distribution. However, it's worth noting that "non-collusion" does not mean "immune to legal compulsion" — in certain jurisdictions, governments may simultaneously issue data disclosure orders to both parties, bypassing this privacy mechanism entirely. Choosing relay and gateway combinations across different jurisdictions is therefore a factor to consider for more serious privacy requirements.
Use Cases and Limitations
OHTTP's most typical use cases include: app telemetry and analytics reporting, password breach detection queries (such as checking whether a password appears in known breach databases), and any scenario where "hiding the request origin while preserving request functionality" is desired. It is particularly well-suited for situations involving primarily one-way reporting where requests don't require returning large amounts of personalized content.
The limitations are equally clear. OHTTP introduces additional encryption and relay hops, which incurs some latency overhead, making it unsuitable for highly interactive, real-time scenarios. Additionally, since request content is visible to the gateway, OHTTP protects the "association between identity and content" — it does not hide content from the gateway itself. If request content inherently contains identifiable information (such as session tokens or unique identifiers), the privacy benefits of OHTTP will be diminished accordingly.
In other words, OHTTP is one tool in the privacy toolbox, not a universal key. Developers need to evaluate their specific data flows to determine whether it genuinely addresses their privacy requirements.
Password breach detection is one of OHTTP's most representative real-world applications, typically combined with techniques such as k-Anonymity or Private Set Intersection (PSI). Take a service like Have I Been Pwned as an example: the client sends only the first few characters of a password hash prefix, the server returns all hashes matching that prefix, and the client performs the comparison locally — this way, the server never learns the specific password being queried. With OHTTP layered on top, the server is further prevented from associating query behavior with the client's IP, protecting even the metadata of "who is querying sensitive passwords." This pattern of combining multiple protocols is a design paradigm worth paying attention to in privacy engineering practice.
Summary
Cloudflare's launch of an OHTTP gateway marks a shift in privacy-enhancing technology — moving from draft standards and proprietary solutions used by a handful of large companies toward more widely accessible managed infrastructure. For developers, this means the cost of building privacy-respecting applications is falling; for the industry as a whole, it pushes privacy protection from an "optional feature" toward a "readily available default capability."
The real test lies in the implementation details: whether the independence of relay and gateway can be guaranteed, whether the latency overhead is acceptable, and whether developers truly understand the boundaries of the trust model. The value of privacy infrastructure ultimately depends on how correctly and thoughtfully it is combined and deployed.
Related articles

Free Unlimited Claude Code Solution: Agnes AI + CC Switch Setup Guide
No paid subscription needed — pair the free AI platform Agnes AI with the open-source tool CC Switch to provide a backend for Claude Code, enabling unlimited free calls to text, image, and video models. This guide covers the full setup, test results, and usage boundaries.

A Complete Zero-Cost Guide to Getting Started with Claude Code for Beginners in China
A beginner's guide to Claude Code for users in China: two installation methods (command line and IDE plugin), full configuration steps for replacing the official API with the free domestic LongCat model, and a hands-on demo of building a small tool with natural language, plus tips.

When the NSA Gets Involved: The AI Supply Chain Crisis Behind Open-Source CVEs and API Key Hijacking
Multiple open-source projects disclosed severe CVEs in one day, PHP flaws alarmed the FBI, and Railway and OpenCode were breached with API keys stolen. This article breaks down the vulnerability window, AI-accelerated exploitation, and the AI supply chain crisis behind the API key black market, with a developer defense checklist.