{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-07-28T06:20:51.496Z","headline":"AI 护栏阻碍进攻性网络安全研究人员的工作","description":"Anthropic 和 OpenAI 等 AI 巨头为限制模型被恶意利用而设置的严格护栏，正妨碍进攻性网络安全研究人员寻找漏洞和开发利用代码。研究人员被迫花费大量时间与模型\"谈判\"而非分析漏洞，或转向无护栏的开源模型。部分研究者因担心数据泄露，仅将云端前沿模型用于逆向工程，而将漏洞发现等核心工作交由本地开源模型完成。","url":"https://www.aioga.com/news/cmrycjow0007iroxsl4i4o7f0/","mainEntityOfPage":"https://www.aioga.com/news/cmrycjow0007iroxsl4i4o7f0/","datePublished":"2026-07-24T01:00:00.000Z","dateModified":"2026-07-24T01:00:00.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://techcrunch.com/2026/07/23/how-ai-guardrails-are-impeding-the-work-of-offensive-cybersecurity-researchers","https://aihot.virxact.com/items/cmrycjow0007iroxsl4i4o7f0"],"canonicalUrl":"https://www.aioga.com/news/cmrycjow0007iroxsl4i4o7f0/","directAnswer":{"@type":"Answer","text":"Aioga 编辑摘要：Anthropic 和 OpenAI 等 AI 巨头为限制模型被恶意利用而设置的严格护栏，正妨碍进攻性网络安全研究人员寻找漏洞和开发利用代码。 Aioga 将其归入「行业动态」方向，重点关注它对真实使用和行业竞争的影响。","url":"https://www.aioga.com/news/cmrycjow0007iroxsl4i4o7f0/","dateCreated":"2026-07-24T01:00:00.000Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"TechCrunch source article","url":"https://techcrunch.com/2026/07/23/how-ai-guardrails-are-impeding-the-work-of-offensive-cybersecurity-researchers","datePublished":"2026-07-24T01:00:00.000Z","provider":{"@type":"Organization","name":"TechCrunch","url":"https://techcrunch.com/2026/07/23/how-ai-guardrails-are-impeding-the-work-of-offensive-cybersecurity-researchers"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmrycjow0007iroxsl4i4o7f0","datePublished":"2026-07-24T01:00:00.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmrycjow0007iroxsl4i4o7f0"}}],"aggregationSource":"TechCrunch：AI（RSS）","originalPublisher":{"name":"TechCrunch","url":"https://techcrunch.com/2026/07/23/how-ai-guardrails-are-impeding-the-work-of-offensive-cybersecurity-researchers"},"article":{"id":"cmrycjow0007iroxsl4i4o7f0","slug":"cmrycjow0007iroxsl4i4o7f0","url":"https://www.aioga.com/news/cmrycjow0007iroxsl4i4o7f0/","title":"AI 护栏阻碍进攻性网络安全研究人员的工作","title_en":"How AI guardrails are impeding the work of offensive cybersecurity researchers","summary":"Anthropic 和 OpenAI 等 AI 巨头为限制模型被恶意利用而设置的严格护栏，正妨碍进攻性网络安全研究人员寻找漏洞和开发利用代码。研究人员被迫花费大量时间与模型\"谈判\"而非分析漏洞，或转向无护栏的开源模型。部分研究者因担心数据泄露，仅将云端前沿模型用于逆向工程，而将漏洞发现等核心工作交由本地开源模型完成。","source":"TechCrunch：AI（RSS）","sourceUrl":"https://techcrunch.com/2026/07/23/how-ai-guardrails-are-impeding-the-work-of-offensive-cybersecurity-researchers","aiHotUrl":"https://aihot.virxact.com/items/cmrycjow0007iroxsl4i4o7f0","publishedAt":"2026-07-24T01:00:00.000Z","category":"行业动态","score":59,"selected":false,"articleBody":["For months, AI giants have devised special vetted programs and strict guardrails to limit the use of their models by malicious hackers. But these limits are now hindering the work of legitimate network defenders, as well as that of offensive cybersecurity researchers.","In June, the U.S. government slapped export control restrictions：https://techcrunch.com/2026/06/12/anthropics-safety-warnings-may-have-just-backfired-the-government-has-pulled-the-plug-on-its-most-powerful-ai/ on Anthropic’s much-hyped AI models Mythos and Fable. The move was prompted at least in part by a report that claimed it was possible to bypass the models’ guardrails designed to prevent users from using them to build and execute malicious cyberattacks.","Regardless of whether the incident was really motivated by fears of a jailbreak：https://techcrunch.com/2026/06/15/the-us-governments-anthropic-models-ban-was-never-about-an-ai-jailbreak/, the fact is that Anthropic has repeatedly marketed：https://techcrunch.com/2026/04/07/anthropic-mythos-ai-model-preview-security/ Mythos as some kind of doomsday cybermachine：https://techcrunch.com/2026/04/09/is-anthropic-limiting-the-release-of-mythos-to-protect-the-internet-or-anthropic/ that can only be given to carefully vetted users, and even then with strict guardrails in place. (The export controls on Fable 5 and Mythos 5 have since been lifted. Fable 5 returned to general access on July 1; Mythos 5 has been reintroduced only to vetted U.S. organizations as part of the government’s review process.)","That kind of gatekeeping isn’t unique to Mythos. Both Anthropic, with its other models, and OpenAI offer cybersecurity researchers programs they can apply to get vetted and — if approved — access models with fewer cybersecurity restrictions: OpenAI’s Trusted Access for Cyber program：https://openai.com/index/scaling-trusted-access-for-cyber-defense/ and Anthropic’s Cyber Verification Program：https://support.claude.com/en/articles/14604842-real-time-cyber-safeguards-on-claude-opus-and-sonnet.","These guardrails have been widely criticized, particularly by researchers whose job is to find unknown vulnerabilities in systems and devise ways to exploit them before criminals do.","During a recent appearance on a cybersecurity podcast, Mark Dowd, a well-known security researcher, said：https://docs.google.com/document/d/1G2B7VetSNfxN9Lfb8f2Y1Vyy-uVBQPl_YSj_3ayVUxM/edit?tab=t.0 that, “it’s not really comfortable to me that these random large companies are making arbitrary decisions about what is safe in security and what’s not.”","Dowd has spent decades finding and selling “zero-days”：https://www.vice.com/en/article/iphone-zero-days-inside-azimuth-security/ — previously unknown software flaws and the exploits that take advantage of them — to Western governments, rather than reporting them to the software makers so they get patched. Governments pay a premium for vulnerabilities precisely because they stay open, which is useful for intelligence operations.","Dowd admitted his work may make him biased, but he isn’t alone. Several people who work in offensive cybersecurity — they proactively probe systems for weaknesses — described to TechCrunch how they use AI tools and deal with their guardrails.","Chris Anley, the chief scientist at security consulting giant NCC Group, said that asking an AI model to try to exploit a bug is a key step in confirming it’s a real vulnerability worth fixing. But if a guardrail prompts the model to refuse to answer the question outright, the guardrail hurts defenders, he said.","“This is where the whole offensive versus defensive and guardrails part comes in, because ‘fix this code’ as a prompt is both an essential mechanism for defense but also a roadmap for finding critical vulnerabilities in the code base,” said Anley. “So at the same time, the same tool is both an offensive tool and a defensive tool, and the two can’t really be unpicked.”","It’s “like a hammer,” he continued. “You can’t build a house without a hammer. It’s definitely a tool but it’s also irreducibly a weapon as well.”","When he and his colleagues run into such a roadblock, they sometimes fall back on open source AI models that come with no guardrails at all.","Contact Us Do you use AI models to find vulnerabilities and develops exploits? We’d love to hear from you. From a non-work device and network, you can contact Lorenzo Franceschi-Bicchierai securely on Signal at +1 917 257 1382, or via Telegram and Keybase @lorenzofb, or email：mailto:lorenzo@techcrunch.com/.：mailto:lorenzo@techcrunch.com/","Paolo Stagno, the chief technology officer at Crowdfense, a well-known company that develops, acquires, and sells unknown vulnerabilities to government agencies, agreed with Dowd, saying AI companies “essentially treat customers like children who need babysitting” with their vetted programs and guardrails.","Stagno said he and his colleagues do use frontier models — but only for reverse engineering. They avoid using AI to help find vulnerabilities or build exploits, he said, because feeding that work into a cloud-based model risks leaking sensitive vulnerability data or having it absorbed into future training runs. For that step, he said, they use open source models run locally, as they do not rely on sharing data outside of the model.","Giuseppe Cali, a security researcher who finds zero-days and develops exploits, said guardrails are not impeding his work. That’s because he doesn’t use AI for offensive work; instead, he uses it for initial reverse engineering, to understand the code he’s analyzing, and to build supporting tools. For that, he said, AI tools can speed up the process and allow him to focus on discovering vulnerabilities.","“I still want to own the actual bug discovery and weaponization myself and that wouldn’t change if all guardrails were lifted tomorrow,” said Cali. “I am jealous of my bugs, and I like this game too much to let models play it for me.”","One researcher at a smartphone-component manufacturer, who spoke on condition of anonymity because he isn’t authorized to talk to the press, said his employer isn’t part of Anthropic’s CVP program and as a result, its tools are barely useful for finding vulnerabilities because the guardrails are too strict.","“If it catches wind we’re doing anything security related, it just stops and isn’t usable,” the person said.","Chris Thompson — chief executive of cybersecurity firm RemoteThreat and founder of Offensive AI Con, an offensive security and AI-focused event — said that in his experience using the frontier AI models, the guardrails can be inconsistent and work differently every day. That’s true even inside the looser boundaries of Anthropic’s and OpenAI’s vetted programs.","“I think the practical impact is you spend a lot of time negotiating with the model instead of working on the core security program,” said Thompson. “Instead of analyzing a vulnerability and reasoning through the exploitability, you’re trying to find why you’re getting inconsistent results or why are models over-sanitizing the output.”","Consequently, researchers rely on or get pushed toward Chinese open source models like GLM — freely downloadable models that can be run locally with no vetting or usage restrictions — said Thompson.","“You have these responsible researchers that are being pushed away from U.S.-governed systems to foreign-owned systems,” he said. “I think it’s more harmful than good to have these guardrails in place.”","Rather than tightening restrictions further, Thompson called for the AI frontier labs to open up their programs, provide responsible access, and hold those who abuse their tools accountable. Otherwise, he argued, defenders will lose the AI race.","“There’s this big storm coming. There’s this big wave of attacks that are going to happen at speed and scale like never before,” said Thompson. “But the same security consulting firms and legit researchers that are trying to make a difference are being stifled right now.”","When you purchase through links in our articles, we may earn a small commission：https://techcrunch.com/techcrunch-affiliate-monetization-standards/. This doesn’t affect our editorial independence.","You can contact or verify outreach from Lorenzo by emailing lorenzo@techcrunch.com：mailto:lorenzo@techcrunch.com, via encrypted message at +1 917 257 1382 on Signal, and @lorenzofb on Keybase/Telegram.","Scale faster. Grow your portfolio. Gain practical expertise. No matter your goal, Disrupt can empower you. Save up to $330 toda y!","SpaceX launches new V3 Starlink satellites but suffers another booster failure：https://techcrunch.com/2026/07/24/spacex-launches-new-v3-starlink-satellites-but-suffers-another-booster-failure/ Sean O'Kane：https://techcrunch.com/author/sean-okane/","US accuses American of allegedly wiping his phone using a ‘duress’ password during border search：https://techcrunch.com/2026/07/24/us-accuses-american-of-allegedly-wiping-his-phone-using-a-duress-password-during-border-search/ Zack Whittaker：https://techcrunch.com/author/zack-whittaker/","Anduril reportedly in talks to raise funding at $100B valuation, more than 3x last year’s mark：https://techcrunch.com/2026/07/24/anduril-reportedly-in-talks-to-raise-funding-at-100b-valuation-more-than-3x-last-years-mark/ Ram Iyer：https://techcrunch.com/author/ram-iyer/","Tesla’s robotaxis are moving in reverse：https://techcrunch.com/2026/07/23/teslas-robotaxis-are-moving-in-reverse/ Sean O'Kane：https://techcrunch.com/author/sean-okane/","Jack Dorsey is taking on Slack with Buzz, a group chat platform for teams and their AI agents：https://techcrunch.com/2026/07/21/jack-dorsey-is-taking-on-slack-with-buzz-a-group-chat-platform-for-teams-and-their-ai-agents/ Amanda Silberling：https://techcrunch.com/author/amanda-silberling/","Light made a flip phone — it’s colorful and it’s cheap：https://techcrunch.com/2026/07/21/light-is-modernizing-the-flip-phone-with-light-flip/ Amanda Silberling：https://techcrunch.com/author/amanda-silberling/","AI music generator Suno breach affects 55M users, per Have I Been Pwned：https://techcrunch.com/2026/07/21/ai-music-generator-suno-breach-affects-55m-users-per-have-i-been-pwned/ Zack Whittaker：https://techcrunch.com/author/zack-whittaker/"],"articleImages":[{"sourceUrl":"https://techcrunch.com/wp-content/uploads/2025/07/Lorenzo-headshot-2023-cropped.jpeg?w=150","alt":"Lorenzo Franceschi-Bicchierai","afterParagraph":25,"url":"/media/articles/cmrycjow0007iroxsl4i4o7f0/57f3c9c7e4a327fe.webp"},{"sourceUrl":"https://techcrunch.com/wp-content/uploads/2025/04/Disrupt2026-Color.png","alt":"Event Logo","afterParagraph":26,"url":"/media/articles/cmrycjow0007iroxsl4i4o7f0/d1d06f6f97361f52.webp"}],"mediaStatus":"ok","articleBodyZh":["几个月来，人工智能巨头们制定了特别审核的程序和严格的防护措施，以限制恶意黑客使用他们的模型。但是这些限制现在正在阻碍合法的网络防御者以及进攻性网络安全研究人员的工作。","今年六月，美国政府对Anthropic备受关注的AI模型Mythos和Fable实施了出口管制限制：https://techcrunch.com/2026/06/12/anthropics-safety-warnings-may-have-just-backfired-the-government-has-pulled-the-plug-on-its-most-powerful-ai/。此举至少部分是由于一份报告声称可能绕过这些模型的防护措施，从而让用户使用它们来构建和执行恶意网络攻击。","无论这一事件是否真的是出于对越狱的担忧：https://techcrunch.com/2026/06/15/the-us-governments-anthropic-models-ban-was-never-about-an-ai-jailbreak/，事实是Anthropic一再将Mythos营销为某种末日网络机器：https://techcrunch.com/2026/04/09/is-anthropic-limiting-the-release-of-mythos-to-protect-the-internet-or-anthropic/，只提供给经过严格审核的用户，即便如此也必须有严格的防护措施。（Fable 5 和 Mythos 5 的出口管控此后已被解除。Fable 5 于7月1日恢复一般访问；Mythos 5 仅在政府审查流程中重新向经过审核的美国组织开放。）","这种守门行为并非Mythos独有。Anthropic的其他模型以及OpenAI都为网络安全研究人员提供了可以申请的项目，如果批准，可访问具有较少网络安全限制的模型：OpenAI的网络防御可信访问计划：https://openai.com/index/scaling-trusted-access-for-cyber-defense/ 以及Anthropic的网络验证计划：https://support.claude.com/en/articles/14604842-real-time-cyber-safeguards-on-claude-opus-and-sonnet。","这些防护措施受到广泛批评，尤其是那些负责发现系统未知漏洞并在犯罪分子之前设计利用方法的研究人员。","在最近一次参加网络安全播客时，知名安全研究员马克·道德（Mark Dowd）说：https://docs.google.com/document/d/1G2B7VetSNfxN9Lfb8f2Y1Vyy-uVBQPl_YSj_3ayVUxM/edit?tab=t.0，“这些随机的大公司对安全性什么是安全、什么不安全做出任意决定让我感到并不舒服。”","道德多年来一直在寻找并出售“零日漏洞” https://www.vice.com/en/article/iphone-zero-days-inside-azimuth-security/——此前未知的软件漏洞以及利用这些漏洞的攻击手段——卖给西方政府，而不是报告给软件制造商让其修补。政府之所以愿意为漏洞支付高价，正是因为漏洞保持开放，这对情报行动很有用。","道德承认他的工作可能让他有偏见，但他并不孤单。几位从事进攻性网络安全的人员——他们主动探查系统弱点——向TechCrunch描述了他们如何使用人工智能工具以及如何处理这些工具的限制措施。","安全咨询巨头NCC集团的首席科学家克里斯·安利（Chris Anley）表示，让AI模型尝试利用漏洞是确认它是否是值得修复的真正漏洞的关键步骤。但如果限制措施导致模型完全拒绝回答问题，那么这些限制会对防御者产生不利影响，他说。","安利说：“这就是整个进攻与防御以及限制措施问题出现的地方，因为‘修复这段代码’作为提示既是防御的必要机制，也是寻找代码库中关键漏洞的路线图。因而同一时间，同一个工具既是进攻工具也是防御工具，这两者实际上无法分开。”","他继续说道：“就像一把锤子。没有锤子就盖不了房子。它确实是一个工具，同时也是不可分割的武器。”","当他和同事遇到这样的阻碍时，他们有时会转而使用完全没有限制措施的开源AI模型。","联系我们 你是否使用 AI 模型来寻找漏洞并开发漏洞利用？我们很想听到你的想法。你可以通过非工作设备和网络，通过 Signal 安全联系 Lorenzo Franceschi-Bicchierai，号码为 +1 917 257 1382，或者通过 Telegram 和 Keybase @lorenzofb，或发送电子邮件：mailto:lorenzo@techcrunch.com/。","Paolo Stagno 是 Crowdfense 的首席技术官，这是一家知名公司，开发、收购并向政府机构出售未知漏洞。他同意 Dowd 的观点，称 AI 公司“本质上把客户当作需要看管的孩子”，提供经过审核的程序和安全防护措施。","Stagno 表示，他和同事确实使用前沿模型——但仅用于逆向工程。他说，他们避免使用 AI 来帮助寻找漏洞或构建漏洞利用工具，因为将这类工作输入云端模型可能会泄露敏感漏洞数据，或被吸收进未来的训练中。对于这一步，他说，他们使用本地运行的开源模型，因为这些模型不依赖于在外部共享数据。","Giuseppe Cali 是一名安全研究员，发现零日漏洞并开发漏洞利用工具。他表示，安全防护措施并未阻碍他的工作。这是因为他不使用 AI 来进行攻击性工作；相反，他使用 AI 进行初步逆向工程，以理解他正在分析的代码，并构建辅助工具。为此，他说，AI 工具可以加快流程，使他能够专注于发现漏洞。","“我仍然希望亲自掌控漏洞发现和武器化的过程，如果明天所有安全防护措施都被取消，我也不会改变这一点，”Cali 说。“我对自己的漏洞心怀占有欲，我太喜欢这场游戏，不想让模型替我玩。”","一家智能手机零部件制造商的一名研究人员表示，由于未获授权接受采访，他要求匿名。他表示，他的雇主不属于 Anthropic 的 CVP 计划，因此其工具在寻找漏洞方面几乎没有用处，因为安全防护措施过于严格。","“如果它察觉到我们在做任何与安全相关的事情，它就会停止工作，无法使用，”该人士说。","Chris Thompson——网络安全公司RemoteThreat的首席执行官，以及以进攻性安全和人工智能为重点的活动Offensive AI Con的创始人——表示，根据他使用前沿AI模型的经验，安全防护措施可能不一致，而且每天的表现都不同。这即使在Anthropic和OpenAI经过审查的项目相对宽松的界限内也是如此。","“我认为实际影响是，你花很多时间与模型进行‘谈判’，而不是专注于核心安全项目，”Thompson说。“你不是在分析漏洞并推理其可利用性，而是在试图找出为什么结果不一致，或者为什么模型过度清理输出。”","因此，研究人员依赖或被推向中国开源模型，如GLM——这些是可以免费下载、在本地运行且没有审查或使用限制的模型——Thompson表示。","“你会看到这些负责任的研究人员被从美国管理的系统推向外国拥有的系统，”他说。“我认为设置这些安全防护措施利大于弊的可能性不大。”","Thompson呼吁，前沿AI实验室应开放他们的项目，提供负责任的访问权限，并追究滥用其工具的人的责任，而不是进一步收紧限制。否则，他认为，防守方将在人工智能竞赛中落后。","“有一场巨大的风暴即将来临。将会有前所未有速度和规模的攻击浪潮，”Thompson说。“但同一批致力于做出改变的安全咨询公司和合法研究人员，现在却被压制。”","当您通过我们文章中的链接购买时，我们可能会赚取少量佣金：https://techcrunch.com/techcrunch-affiliate-monetization-standards/。这不会影响我们的编辑独立性。","您可以通过发送邮件至 lorenzo@techcrunch.com 与Lorenzo联系或确认其联系，或通过Signal加密信息发送至 +1 917 257 1382，以及在Keybase/Telegram上的 @lorenzofb。","更快扩展。增长您的投资组合。获取实际经验。不管您的目标是什么，Disrupt都能赋能您。今天最多可节省330美元！","SpaceX 发射新的 V3 Starlink 卫星，但再次遭遇助推器故障：https://techcrunch.com/2026/07/24/spacex-launches-new-v3-starlink-satellites-but-suffers-another-booster-failure/ Sean O'Kane：https://techcrunch.com/author/sean-okane/","美国指控一名美国人在边境检查时 allegedly 使用‘胁迫’密码擦除手机：https://techcrunch.com/2026/07/24/us-accuses-american-of-allegedly-wiping-his-phone-using-a-duress-password-during-border-search/ Zack Whittaker：https://techcrunch.com/author/zack-whittaker/","据报道，Anduril 正在谈判筹集资金，估值达 1000 亿美元，是去年的三倍多：https://techcrunch.com/2026/07/24/anduril-reportedly-in-talks-to-raise-funding-at-100b-valuation-more-than-3x-last-years-mark/ Ram Iyer：https://techcrunch.com/author/ram-iyer/","特斯拉的机器人出租车正在倒退行驶：https://techcrunch.com/2026/07/23/teslas-robotaxis-are-moving-in-reverse/ Sean O'Kane：https://techcrunch.com/author/sean-okane/","Jack Dorsey 正在用 Buzz 挑战 Slack，这是一款面向团队及其 AI 代理的群聊平台：https://techcrunch.com/2026/07/21/jack-dorsey-is-taking-on-slack-with-buzz-a-group-chat-platform-for-teams-and-their-ai-agents/ Amanda Silberling：https://techcrunch.com/author/amanda-silberling/","Light 发布了一款翻盖手机——它色彩丰富且价格便宜：https://techcrunch.com/2026/07/21/light-is-modernizing-the-flip-phone-with-light-flip/ Amanda Silberling：https://techcrunch.com/author/amanda-silberling/","AI 音乐生成器 Suno 发生数据泄露，影响 5500 万用户，据 Have I Been Pwned 报告：https://techcrunch.com/2026/07/21/ai-music-generator-suno-breach-affects-55m-users-per-have-i-been-pwned/ Zack Whittaker：https://techcrunch.com/author/zack-whittaker/"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"Aioga 编辑摘要：Anthropic 和 OpenAI 等 AI 巨头为限制模型被恶意利用而设置的严格护栏，正妨碍进攻性网络安全研究人员寻找漏洞和开发利用代码。 Aioga 将其归入「行业动态」方向，重点关注它对真实使用和行业竞争的影响。","background":"背景分析：模型与研究类动态需要结合能力边界、开放方式、成本、可用性和真实任务表现判断，单项指标领先不等于已经形成稳定采用。","viewpoint":"Aioga 判断：这条动态更适合作为行业观察信号，当前信息足以建立线索，但不足以推导长期结论。","implications":"影响分析：对相关团队而言，短期应先核对来源、可用范围和实际成本，再判断是否值得接入或跟进。","nextStep":"后续观察：继续观察官方文档、实际可用性、价格变化、开发者反馈和竞品回应。","evidenceRefs":["title","summary","articleBody"],"confidence":"medium","status":"published","aiGenerated":false,"autoApproved":true,"generatedBy":"rule-safe-fallback","generatedAt":"2026-07-28T06:29:10.563Z","sourceHash":"ebc53371f8602b7d","validation":{"passed":true,"mode":"rule-safe-fallback","checks":["schema","length","source-attribution","no-html"]}},"tags":["行业动态","TechCrunch：AI（RSS）"],"translations":{"zh-CN":{"title":"AI 护栏阻碍进攻性网络安全研究人员的工作","summary":"Anthropic 和 OpenAI 等 AI 巨头为限制模型被恶意利用而设置的严格护栏，正妨碍进攻性网络安全研究人员寻找漏洞和开发利用代码。研究人员被迫花费大量时间与模型\"谈判\"而非分析漏洞，或转向无护栏的开源模型。部分研究者因担心数据泄露，仅将云端前沿模型用于逆向工程，而将漏洞发现等核心工作交由本地开源模型完成。","category":"行业动态","source":"TechCrunch","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"AI 护栏阻碍进攻性网络安全研究人员的工作 - Aioga AI资讯","description":"Anthropic 和 OpenAI 等 AI 巨头为限制模型被恶意利用而设置的严格护栏，正妨碍进攻性网络安全研究人员寻找漏洞和开发利用代码。研究人员被迫花费大量时间与模型\"谈判\"而非分析漏洞，或转向无护栏的开源模型。部分研究者因担心数据泄露，仅将云端前沿模型用于逆向工程，而将漏洞发现等核心工作交由本地开源模型完成。","url":"https://www.aioga.com/news/cmrycjow0007iroxsl4i4o7f0/"},"en":{"title":"AI guardrails hinder the work of offensive cybersecurity researchers","summary":"The strict guardrails set by AI giants such as Anthropic and OpenAI to prevent their models from being maliciously used are hindering offensive cybersecurity researchers from finding vulnerabilities and developing exploits. Researchers are forced to spend a lot of time \"negotiating\" with the models instead of analyzing vulnerabilities, or they turn to open-source models without guardrails. Some researchers, out of concern for data leaks, only use cutting-edge cloud models for reverse engineering, while leaving core tasks such as vulnerability discovery to local open-source models.","category":"Industry","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"AI guardrails hinder the work of offensive cybersecurity researchers - Aioga AI News","description":"The strict guardrails set by AI giants such as Anthropic and OpenAI to prevent their models from being maliciously used are hindering offensive cybersecurity researchers from findi...","url":"https://www.aioga.com/en/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:41:46.073Z"},"ja":{"title":"AIのガードレールが攻撃的なサイバーセキュリティ研究者の仕事を妨げる","summary":"AnthropicやOpenAIなどのAI大手が、モデルの悪用を防ぐために設置した厳格なガードレールは、攻撃的なサイバーセキュリティ研究者が脆弱性を見つけたり、エクスプロイトコードを開発したりするのを妨げている。研究者は、脆弱性を分析するのではなくモデルと「交渉」するのに多くの時間を費やさざるを得ず、あるいはガードレールのないオープンソースモデルに頼らざるを得ない。データ漏洩を懸念する一部の研究者は、クラウド上の最先端モデルをリバースエンジニアリングのみに使用し、脆弱性の発見などのコア作業はローカルのオープンソースモデルに任せている。","category":"業界動向","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"AIのガードレールが攻撃的なサイバーセキュリティ研究者の仕事を妨げる - Aioga AIニュース","description":"AnthropicやOpenAIなどのAI大手が、モデルの悪用を防ぐために設置した厳格なガードレールは、攻撃的なサイバーセキュリティ研究者が脆弱性を見つけたり、エクスプロイトコードを開発したりするのを妨げている。研究者は、脆弱性を分析するのではなくモデルと「交渉」するのに多くの時間を費やさざるを得ず、あるいはガードレールのないオープンソースモデルに頼らざるを...","url":"https://www.aioga.com/ja/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:41:59.012Z"},"ko":{"title":"AI 울타리는 공격적인 사이버 보안 연구자들의 작업을 방해한다","summary":"Anthropic와 OpenAI 등 AI 거대 기업들이 모델의 악의적 이용을 제한하기 위해 설정한 엄격한 안전 장치들은 공격적인 사이버 보안 연구자들이 취약점을 찾고 익스플로잇 코드를 개발하는 것을 방해하고 있다. 연구자들은 취약점을 분석하기보다 모델과 '협상'하는 데 많은 시간을 써야 하거나, 안전 장치가 없는 오픈소스 모델로 전환해야 한다. 일부 연구자들은 데이터 유출을 우려해 클라우드 최첨단 모델은 리버스 엔지니어링에만 사용하고, 취약점 발견 등 핵심 작업은 로컬 오픈소스 모델에 맡기고 있다.","category":"업계 동향","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"AI 울타리는 공격적인 사이버 보안 연구자들의 작업을 방해한다 - Aioga AI 뉴스","description":"Anthropic와 OpenAI 등 AI 거대 기업들이 모델의 악의적 이용을 제한하기 위해 설정한 엄격한 안전 장치들은 공격적인 사이버 보안 연구자들이 취약점을 찾고 익스플로잇 코드를 개발하는 것을 방해하고 있다. 연구자들은 취약점을 분석하기보다 모델과 '협상'하는 데 많은 시간을 써야 하거나, 안전 장치가 없는 오픈소스...","url":"https://www.aioga.com/ko/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:42:40.116Z"},"es":{"title":"Las barreras de IA obstaculizan el trabajo de los investigadores de ciberseguridad ofensiva","summary":"Los estrictos límites establecidos por gigantes de la IA como Anthropic y OpenAI para evitar que sus modelos sean utilizados maliciosamente están obstaculizando a los investigadores de seguridad cibernética ofensiva en la búsqueda de vulnerabilidades y el desarrollo de código de explotación. Los investigadores se ven obligados a pasar mucho tiempo \"negociando\" con el modelo en lugar de analizar vulnerabilidades, o bien recurren a modelos de código abierto sin restricciones. Algunos investigadores, preocupados por la fuga de datos, utilizan los modelos avanzados en la nube solo para ingeniería inversa, dejando tareas centrales como la identificación de vulnerabilidades a modelos de código abierto locales.","category":"Industria","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"Las barreras de IA obstaculizan el trabajo de los investigadores de ciberseguridad ofensiva - Aioga Noticias de IA","description":"Los estrictos límites establecidos por gigantes de la IA como Anthropic y OpenAI para evitar que sus modelos sean utilizados maliciosamente están obstaculizando a los investigadore...","url":"https://www.aioga.com/es/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:42:47.648Z"},"fr":{"title":"Les garde-corps de l'IA entravent le travail des chercheurs en cybersécurité offensive","summary":"Les barrières strictes mises en place par les géants de l'IA comme Anthropic et OpenAI pour limiter l'utilisation malveillante des modèles empêchent les chercheurs en cybersécurité offensifs de trouver des vulnérabilités et de développer des codes d'exploitation. Les chercheurs sont contraints de passer beaucoup de temps à \"négocier\" avec les modèles au lieu d'analyser les vulnérabilités, ou de se tourner vers des modèles open source sans barrières. Certains chercheurs, par crainte de fuites de données, n'utilisent les modèles avancés sur le cloud que pour l'ingénierie inverse, confiant la découverte des vulnérabilités et autres travaux essentiels aux modèles open source locaux.","category":"Industrie","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"Les garde-corps de l'IA entravent le travail des chercheurs en cybersécurité offensive - Aioga Actualités IA","description":"Les barrières strictes mises en place par les géants de l'IA comme Anthropic et OpenAI pour limiter l'utilisation malveillante des modèles empêchent les chercheurs en cybersécurité...","url":"https://www.aioga.com/fr/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:43:29.746Z"},"de":{"title":"KI-Schutzvorrichtungen behindern die Arbeit aggressiver Cybersicherheitsforscher","summary":"Die strikten Sicherheitsvorkehrungen, die von KI-Giganten wie Anthropic und OpenAI eingerichtet wurden, um eine böswillige Nutzung ihrer Modelle zu verhindern, behindern derzeit offensiv arbeitende Cybersicherheitsforscher bei der Suche nach Schwachstellen und der Entwicklung von Exploit-Code. Forscher sind gezwungen, viel Zeit damit zu verbringen, mit den Modellen „zu verhandeln“ anstatt Schwachstellen zu analysieren, oder sie wenden sich Open-Source-Modellen ohne Sicherheitsvorkehrungen zu. Einige Forscher nutzen aus Datenschutzgründen die fortschrittlichsten Cloud-Modelle nur für Reverse Engineering, während sie die Kernaufgaben wie die Entdeckung von Schwachstellen lokalen Open-Source-Modellen überlassen.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"KI-Schutzvorrichtungen behindern die Arbeit aggressiver Cybersicherheitsforscher - Aioga KI-News","description":"Die strikten Sicherheitsvorkehrungen, die von KI-Giganten wie Anthropic und OpenAI eingerichtet wurden, um eine böswillige Nutzung ihrer Modelle zu verhindern, behindern derzeit of...","url":"https://www.aioga.com/de/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:43:30.004Z"},"pt-BR":{"title":"Barreiras de IA impedem o trabalho de pesquisadores de segurança cibernética ofensiva","summary":"As barreiras rigorosas estabelecidas por gigantes da IA, como Anthropic e OpenAI, para limitar o uso malicioso dos modelos, estão impedindo os pesquisadores de segurança cibernética ofensiva de encontrar vulnerabilidades e desenvolver códigos de exploração. Os pesquisadores são forçados a gastar muito tempo 'negociando' com os modelos, em vez de analisar vulnerabilidades, ou recorrem a modelos de código aberto sem barreiras. Alguns pesquisadores, preocupados com o vazamento de dados, usam modelos avançados baseados na nuvem apenas para engenharia reversa, deixando a descoberta de vulnerabilidades e outros trabalhos fundamentais para modelos de código aberto locais.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"Barreiras de IA impedem o trabalho de pesquisadores de segurança cibernética ofensiva - Aioga Notícias de IA","description":"As barreiras rigorosas estabelecidas por gigantes da IA, como Anthropic e OpenAI, para limitar o uso malicioso dos modelos, estão impedindo os pesquisadores de segurança cibernétic...","url":"https://www.aioga.com/pt-BR/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:44:05.630Z"},"ru":{"title":"Ограждение ИИ препятствует работе исследователей наступательной кибербезопасности","summary":"Гиганты в области ИИ, такие как Anthropic и OpenAI, устанавливают строгие ограничения, чтобы модели не использовались в преступных целях, что мешает исследователям в области агрессивной кибербезопасности искать уязвимости и разрабатывать эксплойты. Исследователи вынуждены тратить много времени на «переговоры» с моделью вместо анализа уязвимостей или обращаться к открытым моделям без ограничений. Некоторые исследователи из-за опасений утечки данных используют передовые облачные модели только для обратного проектирования, а обнаружение уязвимостей и другую ключевую работу выполняют на локальных открытых моделях.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"Ограждение ИИ препятствует работе исследователей наступательной кибербезопасности - Aioga Новости ИИ","description":"Гиганты в области ИИ, такие как Anthropic и OpenAI, устанавливают строгие ограничения, чтобы модели не использовались в преступных целях, что мешает исследователям в области агресс...","url":"https://www.aioga.com/ru/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:44:12.137Z"},"ar":{"title":"حواجز الذكاء الاصطناعي تعيق عمل باحثي الأمن السيبراني الهجومي","summary":"تقوم شركات الذكاء الاصطناعي الكبرى مثل Anthropic و OpenAI بوضع حواجز صارمة لمنع استغلال النماذج بطريقة ضارة، مما يعيق الباحثين في الأمن السيبراني الهجومي من البحث عن الثغرات وتطوير شفرات الاستغلال. يُضطر الباحثون لقضاء وقت طويل في \"التفاوض\" مع النماذج بدلاً من تحليل الثغرات، أو التوجه إلى النماذج مفتوحة المصدر غير المقيدة. بعض الباحثين، خوفًا من تسرب البيانات، يستخدمون فقط النماذج المتقدمة السحابية للهندسة العكسية، بينما يُترك اكتشاف الثغرات والأعمال الأساسية للنماذج مفتوحة المصدر المحلية.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"حواجز الذكاء الاصطناعي تعيق عمل باحثي الأمن السيبراني الهجومي - Aioga أخبار الذكاء الاصطناعي","description":"تقوم شركات الذكاء الاصطناعي الكبرى مثل Anthropic و OpenAI بوضع حواجز صارمة لمنع استغلال النماذج بطريقة ضارة، مما يعيق الباحثين في الأمن السيبراني الهجومي من البحث عن الثغرات وتطوير...","url":"https://www.aioga.com/ar/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:44:56.924Z"},"hi":{"title":"एआइ बैरियर आक्रामक साइबर सुरक्षा शोधकर्ताओं के काम में बाधा डालता है","summary":"Anthropic और OpenAI जैसे AI दिग्गजों द्वारा मॉडल के दुरुपयोग को रोकने के लिए लगाए गए कड़े सुरक्षा उपाय आक्रामक साइबर सुरक्षा शोधकर्ताओं को कमजोरियों को खोजने और शोषण कोड विकसित करने में बाधा डाल रहे हैं। शोधकर्ताओं को कमजोरियों का विश्लेषण करने के बजाय मॉडल के साथ \"मोलभाव\" करने में बहुत समय व्यतीत करना पड़ता है, या वे बिना सुरक्षा उपाय वाले ओपन-सोर्स मॉडलों की ओर रुख करते हैं। कुछ शोधकर्ता डेटा लीक की चिंता के कारण केवल क्लाउड आधारित उच्च स्तरीय मॉडलों का उपयोग रिवर्स इंजीनियरिंग के लिए करते हैं, जबकि कमजोरियों की खोज जैसी मुख्य भूमिकाएँ स्थानीय ओपन-सोर्स मॉडलों को सौंप देते हैं।","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"एआइ बैरियर आक्रामक साइबर सुरक्षा शोधकर्ताओं के काम में बाधा डालता है - Aioga AI समाचार","description":"Anthropic और OpenAI जैसे AI दिग्गजों द्वारा मॉडल के दुरुपयोग को रोकने के लिए लगाए गए कड़े सुरक्षा उपाय आक्रामक साइबर सुरक्षा शोधकर्ताओं को कमजोरियों को खोजने और शोषण कोड विकसित करन...","url":"https://www.aioga.com/hi/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:44:55.767Z"},"it":{"title":"Le barriere dell'IA ostacolano il lavoro dei ricercatori di sicurezza informatica offensiva","summary":"I rigidi sistemi di sicurezza messi in atto dai colossi dell'IA come Anthropic e OpenAI per limitare l'uso dannoso dei modelli stanno ostacolando i ricercatori di sicurezza informatica offensiva nel trovare vulnerabilità e sviluppare codici di sfruttamento. I ricercatori sono costretti a spendere molto tempo a \"negoziare\" con i modelli anziché analizzare le vulnerabilità, o a rivolgersi a modelli open source senza restrizioni. Alcuni ricercatori, temendo la fuga di dati, utilizzano i modelli all'avanguardia basati su cloud solo per l'ingegneria inversa, mentre scoperta delle vulnerabilità e altri compiti cruciali vengono svolti dai modelli open source locali.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"Le barriere dell'IA ostacolano il lavoro dei ricercatori di sicurezza informatica offensiva - Aioga Notizie IA","description":"I rigidi sistemi di sicurezza messi in atto dai colossi dell'IA come Anthropic e OpenAI per limitare l'uso dannoso dei modelli stanno ostacolando i ricercatori di sicurezza informa...","url":"https://www.aioga.com/it/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:45:39.732Z"},"nl":{"title":"AI-hekken belemmeren het werk van offensieve cybersecurity-onderzoekers","summary":"De strikte veiligheidsmaatregelen die AI-giganten zoals Anthropic en OpenAI hebben ingesteld om te voorkomen dat modellen op kwaadwillige wijze worden gebruikt, belemmeren onderzoekers in offensieve cybersecurity bij het zoeken naar kwetsbaarheden en het ontwikkelen van exploitcode. Onderzoekers worden gedwongen veel tijd te besteden aan 'onderhandelen' met het model in plaats van kwetsbaarheden te analyseren, of wenden zich tot open-sourcemodellen zonder veiligheidsmaatregelen. Sommige onderzoekers gebruiken door bezorgdheid over datalekken alleen geavanceerde cloudmodellen voor reverse engineering, terwijl ze de kernactiviteiten zoals het ontdekken van kwetsbaarheden overlaten aan lokale open-sourcemodellen.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"AI-hekken belemmeren het werk van offensieve cybersecurity-onderzoekers - Aioga AI-nieuws","description":"De strikte veiligheidsmaatregelen die AI-giganten zoals Anthropic en OpenAI hebben ingesteld om te voorkomen dat modellen op kwaadwillige wijze worden gebruikt, belemmeren onderzoe...","url":"https://www.aioga.com/nl/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:45:38.134Z"},"tr":{"title":"Yapay zeka koruma çitleri, saldırgan siber güvenlik araştırmacılarının çalışmalarını engelliyor","summary":"Anthropic ve OpenAI gibi AI devleri, modellerin kötü amaçlarla kullanılmasını sınırlamak için kurulan katı korumalar, saldırgan siber güvenlik araştırmacılarının açıkları bulmasını ve exploit kodları geliştirmesini engelliyor. Araştırmacılar, açıkları analiz etmek yerine modellerle “müzakere” etmek için büyük zaman harcamaya veya korumasız açık kaynak modellerine yönelmeye zorlanıyor. Bazı araştırmacılar veri sızıntısı endişesiyle, yalnızca bulut tabanlı ileri düzey modelleri tersine mühendislik için kullanıyor ve açık bulma gibi temel çalışmaları yerel açık kaynak modelleri üzerinde gerçekleştiriyor.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"Yapay zeka koruma çitleri, saldırgan siber güvenlik araştırmacılarının çalışmalarını engelliyor - Aioga AI Haberleri","description":"Anthropic ve OpenAI gibi AI devleri, modellerin kötü amaçlarla kullanılmasını sınırlamak için kurulan katı korumalar, saldırgan siber güvenlik araştırmacılarının açıkları bulmasını...","url":"https://www.aioga.com/tr/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:46:20.922Z"},"vi":{"title":"Hàng rào AI cản trở công việc của các nhà nghiên cứu bảo mật mạng mang tính tấn công","summary":"Các công ty AI lớn như Anthropic và OpenAI đã thiết lập các biện pháp bảo vệ nghiêm ngặt nhằm hạn chế mô hình bị sử dụng ác ý, đang cản trở các nhà nghiên cứu an ninh mạng tấn công tìm lỗ hổng và phát triển mã khai thác. Các nhà nghiên cứu buộc phải dành nhiều thời gian để 'thương lượng' với mô hình thay vì phân tích lỗ hổng, hoặc chuyển sang các mô hình mở không có biện pháp bảo vệ. Một số nhà nghiên cứu vì lo ngại rò rỉ dữ liệu, chỉ sử dụng các mô hình tiên tiến trên đám mây cho kỹ thuật đảo ngược, trong khi việc phát hiện lỗ hổng và các công việc cốt lõi khác được thực hiện bởi các mô hình mở trên máy local.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"Hàng rào AI cản trở công việc của các nhà nghiên cứu bảo mật mạng mang tính tấn công - Tin tức AI Aioga","description":"Các công ty AI lớn như Anthropic và OpenAI đã thiết lập các biện pháp bảo vệ nghiêm ngặt nhằm hạn chế mô hình bị sử dụng ác ý, đang cản trở các nhà nghiên cứu an ninh mạng tấn công...","url":"https://www.aioga.com/vi/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:46:21.619Z"},"id":{"title":"Pagar pengaman AI menghalangi pekerjaan peneliti keamanan siber ofensif","summary":"Raksasa AI seperti Anthropic dan OpenAI telah menetapkan batasan ketat untuk mencegah model digunakan secara jahat, yang justru menghalangi peneliti keamanan siber ofensif dalam mencari celah dan mengembangkan kode eksploitasi. Peneliti dipaksa menghabiskan banyak waktu untuk 'bernegosiasi' dengan model daripada menganalisis celah, atau beralih ke model open-source tanpa batasan. Beberapa peneliti karena khawatir kebocoran data, hanya menggunakan model cloud canggih untuk rekayasa terbalik, sementara penemuan celah dan pekerjaan inti lainnya diserahkan pada model open-source lokal.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"Pagar pengaman AI menghalangi pekerjaan peneliti keamanan siber ofensif - Berita AI Aioga","description":"Raksasa AI seperti Anthropic dan OpenAI telah menetapkan batasan ketat untuk mencegah model digunakan secara jahat, yang justru menghalangi peneliti keamanan siber ofensif dalam me...","url":"https://www.aioga.com/id/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:47:00.623Z"},"th":{"title":"รั้วป้องกัน AI ขัดขวางงานของนักวิจัยด้านความปลอดภัยไซเบอร์เชิงรุก","summary":"บริษัทยักษ์ใหญ่ด้าน AI เช่น Anthropic และ OpenAI ได้ตั้งมาตรการควบคุมอย่างเข้มงวดเพื่อจำกัดการใช้โมเดลในทางที่เป็นอันตราย ซึ่งขัดขวางนักวิจัยด้านความมั่นคงปลอดภัยทางไซเบอร์ที่มุ่งโจมตีในการค้นหาช่องโหว่และพัฒนารหัสเอ็กซ์พลอยต์ นักวิจัยจึงต้องใช้เวลามากใน การ“เจรจากับ”โมเดล แทนที่จะวิเคราะห์ช่องโหว่ หรือหันไปใช้โมเดลโอเพนซอร์สที่ไม่มีมาตรการควบคุม นักวิจัยบางคนเนื่องจากกังวลเรื่องการรั่วไหลของข้อมูล จึงใช้โมเดลล่าสุดบนคลาวด์เฉพาะสำหรับวิศวกรรมย้อนกลับ ส่วนงานสำคัญอย่างการค้นหาช่องโหว่จะทำผ่านโมเดลโอเพนซอร์สในเครื่องของตนเอง","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"รั้วป้องกัน AI ขัดขวางงานของนักวิจัยด้านความปลอดภัยไซเบอร์เชิงรุก - ข่าว AI Aioga","description":"บริษัทยักษ์ใหญ่ด้าน AI เช่น Anthropic และ OpenAI ได้ตั้งมาตรการควบคุมอย่างเข้มงวดเพื่อจำกัดการใช้โมเดลในทางที่เป็นอันตราย ซึ่งขัดขวางนักวิจัยด้านความมั่นคงปลอดภัยทางไซเบอร์ที่มุ่งโ...","url":"https://www.aioga.com/th/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:47:08.605Z"},"pl":{"title":"Bariera AI utrudnia pracę agresywnych badaczy cyberbezpieczeństwa","summary":"Surowe zabezpieczenia wprowadzone przez gigantów AI, takich jak Anthropic i OpenAI, aby ograniczyć złośliwe wykorzystanie modeli, utrudniają badaczom zajmującym się ofensywnym bezpieczeństwem cybernetycznym poszukiwanie luk i opracowywanie kodu eksploatacyjnego. Badacze są zmuszeni spędzać dużo czasu na \"negocjacjach\" z modelem zamiast analizować luki lub zwracają się ku otwartym modelom pozbawionym zabezpieczeń. Część badaczy, obawiając się wycieku danych, wykorzystuje nowoczesne modele w chmurze wyłącznie do inżynierii odwrotnej, podczas gdy odkrywanie luk i inne kluczowe prace są wykonywane przez lokalne otwarte modele.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"Bariera AI utrudnia pracę agresywnych badaczy cyberbezpieczeństwa - Aioga Wiadomości AI","description":"Surowe zabezpieczenia wprowadzone przez gigantów AI, takich jak Anthropic i OpenAI, aby ograniczyć złośliwe wykorzystanie modeli, utrudniają badaczom zajmującym się ofensywnym bezp...","url":"https://www.aioga.com/pl/news/cmrycjow0007iroxsl4i4o7f0/","contentTranslated":true,"sourceHash":"00a472260d1cdc8a","translatedAt":"2026-07-26T10:47:52.560Z"}}}}