NEAR AI pauses cloud inference after detecting unauthorized access

NEAR AI has hit pause on its cloud inference service after discovering that someone got into its systems without permission.

For a company whose entire sales pitch is confidential AI, this is an awkward week. The firm’s response, though, leans heavily on the same technology it has been selling all along.

What happened at NEAR AI

According to the deep research findings, NEAR AI detected unauthorized access to its internal systems on November 9, 2026. The company suspended NEAR AI Cloud inference operations as a precaution right after spotting the breach.

NEAR AI told users that customer prompts and inference workloads kept running inside hardware-isolated Trusted Execution Environments, or TEEs. The company specifically pointed to Intel TDX and NVIDIA confidential computing as the technologies doing that isolation.

That is the core of NEAR AI’s reassurance. The company says the setup keeps data confidential not only from the cloud provider, but from NEAR AI itself.

By November 10, 2026, NEAR AI confirmed that the unauthorized access had been addressed and closed. An investigation into the breach was still continuing, and the cloud service stayed offline until it could be safely brought back.

The company also said NEAR Protocol and its other NEAR services were unaffected. In other words, the blockchain side of the house kept running while the AI side locked its doors.

The timing is not great

The breach landed shortly after NEAR AI launched a cybersecurity-focused initiative on October 29, 2026. That program was built to offer uncensored models for security research, under the same TEE-based privacy standards the company uses elsewhere.

Little else about the incident has been made public so far. NEAR AI has not described how the access occurred, who was behind it, or what specific systems were touched beyond calling them internal.

Why the TEE argument matters

The interesting wrinkle here is the gap between a company getting breached and customer data getting exposed. NEAR AI is arguing they are not the same thing.

If that design held up as NEAR AI says it did, this incident becomes a live test of the confidential computing model. An intruder in the company’s systems would, in theory, still be locked out of the prompts and workloads running inside the enclaves.

That is a meaningful claim, but it is the company’s claim. The investigation is still open, and independent confirmation of what the intruder could and could not see has not been published.

What this means for NEAR AI and the NEAR ecosystem

The research material flags that investors and traders may view the breach as a question mark over platforms handling sensitive data. Because NEAR AI’s value proposition rests on TEE-based privacy, any negative fallout could dent its reputation and potentially spill over to the broader NEAR ecosystem.

The company has tried to wall that off by stating that NEAR Protocol and other NEAR services were not affected.

There is also a broader regulatory dimension. The research notes that investors may want to watch for potential long-term regulatory scrutiny in the tech space as the investigation plays out.