东京 —— 据《日经新闻》获悉，日本政府已呼吁美国人工智能开发商Anthropic和OpenAI提供更多使用先进模型的权限，这些模型用于发现网络安全漏洞，并将鼓励关键基础设施运营商使用这些模型。
TOKYO -- The Japanese government has called on U.S. artificial intelligence developers Anthropic and OpenAI to provide greater access to advanced models used to find cybersecurity weaknesses and will encourage their use by operators of essential infrastructure, Nikkei has learned.
政府计划在最早下周出台的网络安全方案中，纳入与美国AI开发商加强合作的内容。
The government plans to include stronger cooperation with U.S. AI developers in a cybersecurity package due out as early as next week.
首相高市早苗周一要求内阁成员推动相关工作，与海外当局合作识别攻击者并解决漏洞，因为日本正努力遏制一波数据泄露浪潮。
Prime Minister Sanae Takaichi called on cabinet members Monday to promote efforts to identify attackers and address vulnerabilities in cooperation with overseas authorities as Japan struggles to contain a wave of data breaches.
网络安全担当大臣古川俊治10月13日与Anthropic联合创始人杰克·克拉克会面，确认该公司与政府的AI安全研究所将合作分析AI赋能的攻击手段。
Cybersecurity Minister Toshiharu Furukawa met with Anthropic co-founder Jack Clark on Oct. 13, confirming the company and the government's AI Safety Institute would work together to analyze AI-enabled attack methods.
政府与执政党自由民主党代表上周与OpenAI高管会面，讨论在应对AI相关漏洞方面的合作，例如近期曝光的编程语言PHP中的缺陷。
Representatives from the government and the ruling Liberal Democratic Party met with OpenAI executives last week, discussing cooperation on responding to AI-related vulnerabilities, such as the recently exposed faults in the programming language PHP.
东京计划使用先进AI模型并比较其性能和特点，以更好地保护处理个人信息或关乎国家安全的关键平台。
Tokyo plans to use advanced AI models and compare their performance and characteristics to better safeguard critical platforms that handle personal information or are linked to national security.
官方认为，近期针对日本企业及其他组织的接连网络攻击，可能涉及使用了先进AI，使黑客能比单靠人力更快、大规模得多地渗透系统。尖端AI能在黑客利用漏洞之前发现并修复它们。
Officials believe the spate of recent cyberattacks on Japanese companies and other organizations may have involved the use of advanced AI that enables hackers to penetrate systems faster and on a far larger scale than humans could alone. Cutting-edge AI can find and fix vulnerabilities before hackers can exploit them.
6月，政府表示已获得使用Anthropic强大的Claude Mythos的权限。随后，政府获得了该公司先进AI模型的免费使用额度，并表示一直在用它们评估政府系统及部分私营企业的漏洞。
In June, the government said it had obtained access to Anthropic's powerful Claude Mythos It later secured a free usage allocation for the company's advanced AI models and said it has been using them to assess vulnerabilities in government systems and some private-sector companies.
东京还获得了使用OpenAI的GPT-5.5-Cyber的权限。
Tokyo has also obtained access to OpenAI's GPT-5.5-Cyber.
近几个月来，日本的网络攻击有所增加。据总部位于日本的网络安全云公司称，9月份检测到的针对网站的攻击（包括未经授权的访问尝试）达到约4.5亿次，比8月份增加了约36%。
Cyberattacks in Japan have ramped up in recent months. Detected attacks on websites, including unauthorized access attempts, reached about 450 million in September, up roughly 36% from August, according to Japan-based Cyber Security Cloud.
政府打算借鉴与人工智能开发商合作所获得的经验，以加强私营部门的网络安全，东京一直在敦促私营部门加强防御。
The government intends to apply lessons gained by working with the AI developers to strengthen cybersecurity in the private sector, which Tokyo has been pressing to bolster defenses.
今年5月，政府宣布了一揽子措施，呼吁应用软件补丁来修复漏洞，并建立包括高级管理层在内的更强有力的风险管理框架。这些措施尤其针对电力、燃气和电信公司等关键基础设施的运营商。针对中小型企业的攻击也有所增加。
In May, it announced a package of measures calling for the application of software patches to address vulnerabilities and stronger risk-management frameworks that include senior management. The measures are aimed particularly at operators of critical infrastructure like power, gas and telecommunications companies. Attacks targeting small and midsize businesses have also increased.
一位政府高级官员表示：“修复漏洞当然很重要，但企业首先需要实施基本措施，比如保持软件更新。”
"Addressing vulnerabilities is important, of course, but companies first need to implement basic measures like keeping their software up to date," a senior government official said.