<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"
        xmlns:news="http://www.google.com/schemas/sitemap-news/0.9">
    <url>
        <loc>https://www.winzheng.com/article/ai-doom-warnings-debate</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-14T04:23:43+08:00</news:publication_date>
            <news:title>AI末日警告再起：恐慌背后藏着什么？</news:title>
                        <news:keywords>AI安全, AI监管, 末日论, 行业分析</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/obama-democrats-ai-safeguards-plan</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-14T02:23:58+08:00</news:publication_date>
            <news:title>奥巴马喊话民主党：AI监管必须有“清晰计划”</news:title>
                        <news:keywords>奥巴马, 人工智能监管, 美国政治, AI安全, 民主党</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/ai-agents-power-data-center-boom</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-13T19:23:53+08:00</news:publication_date>
            <news:title>AI智能体“狂饮”电力，数据中心扩建潮席卷硅谷</news:title>
                        <news:keywords>AI智能体, 数据中心, 能源消耗, 硅谷, 算力</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/paul-christiano-joins-openai-foundation-board-safety-committ</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-13T14:26:22+08:00</news:publication_date>
            <news:title>RLHF奠基人保罗·克里斯蒂亚诺加入OpenAI理事会：最强批评者入局，评测独立性存结构性张力</news:title>
                        <news:keywords>AI安全, OpenAI, AI治理, RLHF, 对齐研究, METR, Safety Committee</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/amodei-pace-the-frontier-ai-slowdown-third-party-evaluation</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-13T14:21:24+08:00</news:publication_date>
            <news:title>Amodei呼吁AI前沿减速并承诺第三方永久驻场：三巨头罕见共识背后的真实逻辑</news:title>
                        <news:keywords>Anthropic, 人工智能安全, Dario Amodei, AI监管, OpenAI, 前沿模型治理</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/us-china-ai-safety-dialogue-mid-september-beijing</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-13T13:00:17+08:00</news:publication_date>
            <news:title>美中9月中旬北京首谈AI安全 贝森特领军中方曾提定义前提</news:title>
                        <news:keywords>AI安全, 美中对话, 特朗普政府</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/ai-agents-papercut-breach-395-orgs-off-script</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-13T09:15:21+08:00</news:publication_date>
            <news:title>AI代理违背指令攻破395机构 26秒内11家同时沦陷</news:title>
                        <news:keywords>AI安全, 网络攻击, PaperCut漏洞</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/real-swe-benchmark-fable-5-1-private-codebase-performance</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-13T09:00:22+08:00</news:publication_date>
            <news:title>Real-SWE基准曝光：Fable 5.1私有代码解决率仅38.8%</news:title>
                        <news:keywords>AI评测, 代码生成, 行业基准</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/openai-altman-ipo-2026</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-13T06:23:53+08:00</news:publication_date>
            <news:title>OpenAI IPO急刹车？奥特曼称2026年上市“不明智”</news:title>
                        <news:keywords>OpenAI, 萨姆·奥特曼, IPO, 人工智能, 科技公司上市</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/anthropic-cybersecurity-alignment-threat-intelligence-2026</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-13T06:16:30+08:00</news:publication_date>
            <news:title>Anthropic双报告风暴：四起AI入侵真实系统，154页滥用档案揭武器开发与无人机自主杀伤</news:title>
                        <news:keywords>Anthropic, Claude, AI安全, 网络安全, 威胁情报, AI滥用</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/openai-agents-api-public-beta-codex-infrastructure-cost-anal</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-13T06:10:24+08:00</news:publication_date>
            <news:title>OpenAI托管代理API公测上线，基础设施外包背后的真实成本待检验</news:title>
                        <news:keywords>OpenAI, Agents API, AI代理, Codex, 开发者工具, 云计算, AI基础设施</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/anthropic-pace-the-frontier-plan</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-13T00:23:42+08:00</news:publication_date>
            <news:title>Anthropic CEO：AI竞赛该「调控前沿」而非狂飙</news:title>
                        <news:keywords>Anthropic, AI安全, 前沿模型, AI治理, 达里奥·阿莫代伊</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/openai-pauses-training-altman-coordinates-slowdown</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T22:00:16+08:00</news:publication_date>
            <news:title>OpenAI暂停多项训练任务 阿尔特曼首次承认愿与对手协调减速</news:title>
                        <news:keywords>OpenAI, AI安全, Sam Altman</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/openai-congress-mandatory-ai-safety-regulation-rogue-agents-</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T20:23:16+08:00</news:publication_date>
            <news:title>OpenAI呼吁国会强制立法：AI失控事件倒逼监管从自愿走向强制</news:title>
                        <news:keywords>OpenAI, AI监管, 美国立法, AI安全, 加州法案, AI agent</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/eu-ai-act-enforcement-september-2026-high-risk-audit</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T20:19:38+08:00</news:publication_date>
            <news:title>欧盟AI执法正式开枪：三国监管机构九月入场，合规窗口已关闭</news:title>
                        <news:keywords>欧盟AI法案, AI合规, GPAI, 监管执法, 人工智能风险</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/claude-misuse-ai-security-threats</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T19:24:25+08:00</news:publication_date>
            <news:title>从黑客攻击到生物武器：Claude滥用已成全球威胁</news:title>
                        <news:keywords>AI安全, Claude滥用, 网络安全, 勒索软件, 内容治理</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/unitree-robot-dog-review</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T19:23:45+08:00</news:publication_date>
            <news:title>我花了4000美元，买了一台中国机器狗</news:title>
                        <news:keywords>宇树科技, 机器狗, 四足机器人, 人形机器人</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/anthropic-threat-intelligence-report-september-2026-bioweapo</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T14:26:05+08:00</news:publication_date>
            <news:title>Anthropic发布154页威胁报告：Claude曾被用于生物武器研究，七家中国AI公司被指蒸馏模型逾1.5亿次</news:title>
                        <news:keywords>Anthropic, Claude, AI安全, 生物武器, 模型蒸馏, 中国AI, 地缘政治, AI监管</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/openai-agents-api-gpt-live-1-financial-services-launch</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T14:22:51+08:00</news:publication_date>
            <news:title>OpenAI一日三弹：代理API、语音模型、金融版同日落地，数据主权成唯一短板</news:title>
                        <news:keywords>OpenAI, Agents API, GPT-Live-1, 语音模型, 企业AI, 金融AI, 大模型</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/mecka-ai-500m-valuation-sequoia</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T10:23:51+08:00</news:publication_date>
            <news:title>红杉领投，Mecka AI估值近5亿美元押注机器人训练数据</news:title>
                        <news:keywords>机器人训练数据, 红杉资本, 具身智能, AI融资, Mecka AI</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/techcrunch-disrupt-2026-side-events-deadline</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T09:23:53+08:00</news:publication_date>
            <news:title>最后24小时：Disrupt 2026边会申请今晚截止</news:title>
                        <news:keywords>TechCrunch Disrupt, 科技大会, 创业生态, 边会活动, 风险投资</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/openai-agents-rubygems-attack-may-2026</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T08:45:14+08:00</news:publication_date>
            <news:title>OpenAI智能体5月入侵RubyGems 上传2000余恶意包未提前通知</news:title>
                        <news:keywords>OpenAI, AI安全, RubyGems, 奖励黑客, 开源供应链</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/ai-extinction-risk-roundtable</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T06:26:31+08:00</news:publication_date>
            <news:title>AI末日危机：顶尖实验室的警告是真是假？</news:title>
                        <news:keywords>AI安全, 人工智能风险, AI治理, 存在性风险, MIT科技评论</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/techcrunch-disrupt-2026-exhibit-deadline</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T06:25:42+08:00</news:publication_date>
            <news:title>Disrupt 2026展位申请倒计时：仅剩一周</news:title>
                        <news:keywords>TechCrunch Disrupt, 科技展会, 创业融资, 创投生态, AI创业</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/openai-mathematicians-open-letter</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T06:24:47+08:00</news:publication_date>
            <news:title>OpenAI与数学界矛盾升级：25位数学家联名抗议</news:title>
                        <news:keywords>OpenAI, 数学界抗议, AI训练数据, 知识产权, 学术伦理</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/yc-garry-tan-open-weight-distillation</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T06:23:54+08:00</news:publication_date>
            <news:title>YC的Garry Tan：开源AI实验室也应能蒸馏前沿模型</news:title>
                        <news:keywords>开源AI, 模型蒸馏, Y Combinator, 中美AI竞争, AI政策</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/openai-agents-api-public-beta-codex-infrastructure-september</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T06:20:22+08:00</news:publication_date>
            <news:title>OpenAI推出Agents API公测：把Codex底层基础设施开放给全体开发者，数据暂限美国境内</news:title>
                        <news:keywords>OpenAI, Agents API, AI智能体, 企业级AI, Codex</news:keywords>
                    </news:news>
    </url>
    <url>
        <loc>https://www.winzheng.com/article/anthropic-jacob-coxon-resignation-ai-extinction-risk-evan-hu</loc>
        <news:news>
            <news:publication>
                <news:name>赢政天下</news:name>
                <news:language>zh</news:language>
            </news:publication>
            <news:publication_date>2026-09-12T06:13:33+08:00</news:publication_date>
            <news:title>Anthropic研究员弃权益辞职：内部对齐负责人公开承认AI致人类灭绝概率超10%</news:title>
                        <news:keywords>AI安全, Anthropic, Jacob Coxon, 超级智能, AI监管, Evan Hubinger, OpenAI, AI风险</news:keywords>
                    </news:news>
    </url>
</urlset>
