NEAR AI suspended its Cloud inference service as a precaution after detecting unauthorized access to internal systems on November 9, 2026. The company said customer prompts and inference workloads continued running in hardware-isolated Trusted Execution Environments using Intel TDX and NVIDIA confidential computing, which it says protect data from both the cloud provider and NEAR AI itself. NEAR AI said the access was addressed and closed on November 10, while its investigation continued and the service remained offline pending a safe restart. NEAR Protocol and other NEAR services were unaffected. The incident followed an October 29, 2026, launch of a cybersecurity-focused initiative offering uncensored models under the same TEE-based privacy standards. NEAR AI has not disclosed how the access occurred, who was responsible, or which internal systems were accessed, and independent confirmation of the company’s claims about data exposure has not been published. The breach could test confidence in NEAR AI’s confidential-computing model and draw longer-term regulatory scrutiny, with potential reputational implications for the broader NEAR ecosystem.