英国将加强网络防御,AI 国务大臣呼吁人工智能企业加大安全保障力度UK to step up cyber defences, AI Secretary of State calls for AI companies to step up security


AI free token/AI免费积分
英国将加强网络防御,AI 国务大臣呼吁人工智能企业加大安全保障力度

IT之家 10 月 12 日消息,据《卫报》报道,英国将加强其网络防御,同时呼吁顶尖人工智能研发企业在平台安全治理上采取更积极的举措。

英国 AI 国务大臣卡尼什卡 · 纳拉扬(Kanishka Narayan)预计将在西伦敦发表的专题演讲中正式宣布上述政策主张。

纳拉扬将呼吁前沿人工智能研发企业采取更多切实举措保障平台安全。IT之家获悉,纳拉扬表示,英国绝不能完全依赖科技巨头对其模型开展“自我监管”,并强调英国本土必须构建更多自主可控的 AI 基础设施。

延伸阅读

据英国政府官网资料,卡尼什卡 · 纳拉扬于 2026 年 7 月 20 日被任命为人工智能国务大臣,同时在英国内阁办公室以及商业、创新、科学与贸易部(BIST)任职。他此前于 2025 年 9 月 7 日至 2026 年 7 月 20 日担任科学、创新与工艺部(DSIT)的议会副国务大臣,并于 2024 年 7 月当选威尔士格拉摩根谷选区议员。

纳拉扬分管的内阁办公室职责中涵盖人工智能安全研究院(AI Security Institute,AISI)。该机构旨在构建人工智能风险的证据基础,为政府决策给出参考。据英国议会 2026 年 4 月披露的信息,AISI 已与 OpenAI、谷歌 DeepMind、Anthropic 等领先 AI 公司合作测试了超过 30 个模型,其发现会在模型部署前促成实际改动。

2025 年 12 月 18 日,AISI 发布首份《前沿人工智能趋势报告》。报告表明 AI 平台能力提高迅速:在网络安全领域,模型完成学徒级任务的成功率从 2023 年的不足 9% 升至 2025 年的约 50%,2025 年首次有模型完成需要长达 10 年经验的专家级网络任务;在软件工程领域,模型完成时长约 1 小时的编程任务的成功率从两年前的不足 5% 提高至超过 40%;部分网络任务中,AI 无需人工指导即可完成的时长大约每 8 个月翻一番。报告同时指出,用于确保 AI 按预期运行的防护措施正在改善,红队找到“通用越狱”方法所需时间从几分钟延长至数小时,约提高 40 倍。

在强化本土 AI 基础设施维度,英国政府此前已推出一系列举措:设立规模 5 亿英镑的主权 AI(Sovereign AI)风投基金,首笔股权投资投向 AI 基础设施公司 Callosum;发布《AI 硬件计划》,扶持研发 AI 芯片与半导体工艺的英国企业;发布《网络增长行动计划》(Cyber Growth Action Plan)以提振本国网络安全产业,并扩大国家安全战略投资基金规模。英国政府认为,绝大多数 AI 平台应在使用环节受到监管。

本文来源:https://www.ithome.com/1/011/727.htm
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。

IT House reported on October 12 that the UK will strengthen its cyber defences, while calling on top AI R&D companies to take a more proactive approach to platform security governance.

UK AI Secretary of State Kanishka Narayan is expected to formally announce the policy proposition in a keynote speech in West London.

Narayan will call on cutting-edge AI R&D companies to take more practical measures to ensure platform security. IT House was informed that Narayan said that the UK must not rely solely on tech giants to “self-regulate” its models, and stressed that the UK must build more autonomous and controllable AI infrastructure.

Kanishka Narayan was appointed Minister of State for Artificial Intelligence on 20 July 2026, according to the UK Government's website, while serving in the UK Cabinet Office and in the Department for Business, Innovation, Science and Trade (BIST). He previously served as Parliamentary Under-Secretary of State for Science, Innovation and Technology (DSIT) from 7 September 2025 to 20 July 2026 and was elected to the Glamorgan Valley Welsh Assembly in July 2024.

The responsibilities of the Cabinet Office under Narayan include the AI Security Institute (AISI). The agency aims to build an evidence base on AI risks to inform government decisions. According to information disclosed by the British Parliament in April 2026, AISI has collaborated with leading AI companies such as OpenAI, Google DeepMind, and Anthropic to test more than 30 models, and its findings will lead to actual changes before the models are deployed.

UK to step up cyber defences, AI Secretary of State calls for AI companies to step up security

On December 18, 2025, AISI released the first Frontier Artificial Intelligence Trends Report. The report shows that AI platform capabilities have improved rapidly: in the field of cybersecurity, the success rate of models completing apprentice-level tasks has increased from less than 9% in 2023 to about 50% in 2025, and the first model to complete expert-level network tasks that require up to 10 years of experience in 2025; in the field of software engineering, the success rate of models completing programming tasks that take about 1 hour has increased from less than 5% two years ago to more than 40%; in some network tasks, the length of AI without manual guidance has doubled every 8 months. The report also notes that safeguards to ensure AI is operating as expected are improving, with the time it takes the Red Team to find a “universal jailbreak” approach increasing from minutes to hours, about 40-fold.

In strengthening the dimension of local AI infrastructure, the UK government has launched a series of initiatives: the establishment of a £500 million sovereign AI (Sovereign AI) venture fund, with the first equity investment in Callosum, an AI infrastructure company; the release of the AI Hardware Plan to support UK companies researching and developing AI chips and semiconductor processes; and the release of the Cyber Growth Action Plan to boost the country's cybersecurity industry and expand the size of the National Security Strategic Investment Fund. The UK government believes that the vast majority of AI platforms should be regulated at the point of use.

This article is from: https://www.ithome.com/1/011/727.htm For copyright issues, please write to: 13951431913#139.com or submit a review on the right.

暂无评论,快来发表第一条评论吧!No comments yet. Be the first!

📮 需求咨询📮 Inquiry