东京 -- 据日经获悉，NTT与软银公司正分别扩大部署在日本国内完整处理人工智能数据的网络安全系统，通过在不将数据发送至美国的情况下诊断软件漏洞，助力满足数据主权需求。
TOKYO -- NTT and SoftBank Corp. are separately expanding the rollout of cybersecurity systems that process artificial intelligence data entirely within Japan, helping meet demand for data sovereignty by diagnosing software vulnerabilities without sending data to the U.S., Nikkei has learned.
利用先进AI查找软件漏洞，通常需要将软件源代码发送给AI模型的开发商。由于日本使用的许多最先进AI模型均由美国开发，部分企业担心相关数据流向海外的风险。
Finding software bugs via advanced AI typically entails sending the software's source code to the developer of the AI model. Because many of the most advanced AI models used in Japan are developed in the U.S., some companies are concerned about the risk of related data flowing overseas.
NTT旗下子公司NTT数据集团正在扩大一项无需将源代码发往海外即可使用的新型网络安全服务。该服务采用了与日本初创企业Aeye Security Lab共同开发的系统，可使用多种AI模型。
Through subsidiary NTT Data Group, NTT is expanding a new cybersecurity service that can be used without sending source code abroad. The service employs a system, developed with Japanese startup Aeye Security Lab, that allows the use of multiple AI models.
该系统主要依赖开放AI模型，以便在日本国内管理数据。AI发现软件漏洞后，NTT数据的网络安全专家会对结果进行审查并提出修复方案。
The system relies mainly on open AI models so that data can be managed in Japan. After the AI finds software vulnerabilities, NTT Data cybersecurity specialists review the findings and propose fixes.
随着更多可在日本国内完全部署的AI模型问世，NTT计划陆续引入，包括美国Anthropic和OpenAI的模型。该公司也在关注专为网络安全设计的更先进AI模型。
NTT plans to add more AI models that can be deployed entirely within Japan as they become available, including ones from U.S.-based Anthropic and OpenAI. The company is also eyeing more advanced AI models specially designed for cybersecurity.
该服务于7月推出，面向日本关键基础设施运营商及其他组织进行推广，包括国内金融机构和制造商。
The service was launched in July and is being marketed to operators of critical infrastructure and other organizations in Japan, including domestic financial institutions and manufacturers.
软银目标在未来一至两年内，与OpenAI建立数据完全在日本国内处理的网络安全服务。双方已开始就创建可在日本独家运行OpenAI最先进网络安全专用模型的环境展开讨论。
Within the next one to two years, SoftBank also aims to establish a cybersecurity service with OpenAI where data is processed entirely within Japan. The companies have begun discussions on creating an environment in which OpenAI's most advanced cybersecurity-focused models can be operated exclusively in Japan.
软银于7月开始全面部署基于OpenAI新款GPT-5.5 Cyber模型的网络安全服务。包括金融和电力行业企业在内，已有超过3500家企业报名签约。
SoftBank began full-scale deployment in July of a cybersecurity service based on OpenAI's new GPT-5.5 Cyber model. More than 3,500 enterprises, including companies in the financial and power sectors, have signed up.
尽管该系统目前要求将与客户软件相关的信息首先发送至美国，但软银的目标是未来在日本境内完成所有数据处理。
Although the system currently requires information related to a customer's software to first be sent to the U.S., SoftBank aims to later have all data processed in Japan.