OpenRouter Removes Three Inference Providers Following Data Privacy Concerns

Share

- Advertisement -

OpenRouter has removed three inference providers from its platform and suspended Zero Data Retention (ZDR) eligibility for 125 endpoints following security concerns raised by researchers. The investigation found evidence that certain AI inference providers may have mishandled customer data, raising fresh questions about privacy, transparency and security across the AI supply chain.

The announcement, dated November 22, 2026, follows research highlighting potential weaknesses in how some inference providers process and handle sensitive information. OpenRouter said it had been working directly with the researchers and affected providers to investigate the findings, understand the potential impact and push for stronger data protection measures.

The company stressed that the investigation did not uncover a breach or unauthorised access involving OpenRouter itself. Instead, the findings focused on specific inference providers operating within the wider AI ecosystem.

The incident highlights a growing challenge for businesses and developers using third party AI services. While a platform may offer access to hundreds of models through a single interface, the actual processing of prompts and responses can involve multiple providers. This creates additional security considerations, particularly when requests contain confidential business information, personal details or proprietary data.

Security Research Raises Questions About AI Data Privacy

The concerns emerged after security researchers published findings suggesting that certain inference providers had experienced data exposure or failed to follow their stated data handling policies.

The research examined how providers manage information sent through AI inference services. In particular, it highlighted situations where the handling of prompts and other request data appeared inconsistent with the privacy protections customers expected.

- Advertisement -

For organisations relying on AI tools, this is a significant concern. Prompts sent to an AI model can contain much more than simple questions. Developers may submit source code, internal documents, customer information, business strategies and other sensitive material.

If an inference provider does not follow its declared retention and processing policies, that information could be exposed to risks beyond what customers originally anticipated.

OpenRouter said it had been collaborating with researchers and providers to understand the findings. The company also acknowledged the importance of the research in identifying weaknesses that may otherwise have gone unnoticed.

Importantly, the announcement does not suggest that every inference provider is affected. The investigation identified concerns involving specific providers and endpoints rather than the entire AI infrastructure.

OpenRouter also clarified that there was no evidence of a security breach or unauthorised access to its own systems as part of the reported testing.

OpenRouter Removes Three Providers and 125 Endpoints

Following its investigation, OpenRouter took several measures to reduce potential risks across its provider network.

- Advertisement -

The most significant action was removing three inference providers from the platform. The company also removed 125 endpoints from its Zero Data Retention eligibility programme.

Zero Data Retention is an important privacy consideration for organisations using third party AI services. Under a ZDR arrangement, eligible providers are expected not to retain customer request data beyond the permitted processing requirements.

However, ZDR eligibility depends on the provider’s ability to meet the relevant data handling commitments. If an endpoint cannot provide sufficient assurances, its eligibility becomes a security concern.

OpenRouter said the 125 endpoint removals included providers identified by the researchers, along with additional endpoints removed as a precaution.

This broader approach suggests that the company is not limiting its response to the providers directly named in the investigation. Instead, it is reviewing other endpoints that could present similar risks.

The company has also notified all potentially affected customers.

- Advertisement -

For businesses and developers, this notification is particularly important because the impact of an incident depends on which provider and endpoint handled their requests. Customers who used affected services may need to review their activity and reassess the information they previously submitted.

OpenRouter has not suggested that every customer experienced data exposure. The notifications are intended to inform potentially affected users about the findings and the company’s response.

Network Checks and Stronger Provider Verification Are Underway

Removing providers and suspending endpoint eligibility are immediate steps. However, OpenRouter says additional security work is already underway to strengthen its long term approach.

One priority is adapting the researchers’ testing methods so similar checks can be performed continuously.

This could help identify questionable data handling practices earlier, rather than relying entirely on reports from external researchers or incidents disclosed by providers.

OpenRouter is also conducting technical verification of network paths. The objective is to identify situations in which customer requests and responses might be forwarded to third parties.

This is an important part of the investigation because data protection does not depend solely on a provider’s written privacy policy. It also depends on how information moves through its infrastructure.

For example, a request may pass through multiple systems before reaching the intended model. If another service receives or processes that information, customers need to understand whether the additional processing is covered by the original provider’s privacy commitments.

OpenRouter wants greater visibility into these network connections so it can better assess where customer information travels.

The company is also requiring appropriate assurances and technical validation before restoring ZDR eligibility to any endpoint removed during the review.

This means providers will need to demonstrate that they meet the required standards rather than simply claim compliance.

For customers, this could offer greater confidence when selecting AI models and inference providers through the platform.

Providers Publish Security Updates Following the Investigation

OpenRouter has identified three providers that have issued public security announcements related to the findings.

These announcements provide additional information about the incidents and the providers’ responses.

Novita

Novita published a privacy notice concerning MiniMax requests processed through its infrastructure. The announcement addresses privacy considerations associated with those requests and provides further context about the reported concerns.

Phala

Phala released a preliminary incident report in November. The report outlines its initial findings and provides an update on its investigation into the security concerns.

Together AI

Together AI disclosed a security event involving prompt leakage affecting requests processed during June and July 2026. The incident raises questions about how prompts were handled and the potential exposure of information submitted during that period.

OpenRouter noted that these are not necessarily the only providers involved. Some affected providers have not published public incident reports, although the platform says it has notified customers who may have been affected.

The company is encouraging all providers to publish detailed incident write ups and intends to update its public announcement as additional information becomes available.

This is an important step towards improving transparency across the AI infrastructure market. Customers need clear information about what happened, which services were involved and what measures have been introduced to prevent similar incidents.

Follow TechBSB For More Updates

- Advertisement -
Emily Parker
Emily Parker
Emily Parker is a seasoned tech consultant with a proven track record of delivering innovative solutions to clients across various industries. With a deep understanding of emerging technologies and their practical applications, Emily excels in guiding businesses through digital transformation initiatives. Her expertise lies in leveraging data analytics, cloud computing, and cybersecurity to optimize processes, drive efficiency, and enhance overall business performance. Known for her strategic vision and collaborative approach, Emily works closely with stakeholders to identify opportunities and implement tailored solutions that meet the unique needs of each organization. As a trusted advisor, she is committed to staying ahead of industry trends and empowering clients to embrace technological advancements for sustainable growth.

Read More

Trending Now