切断双向：AI加强系统保护，同时为恶意行为者开辟新的利用。

切断双向：AI加强系统保护，同时为恶意行为者开辟新的利用。

网络安全产业发现强大的前沿AI模型是一把双刃剑。

公司可以比以往更快地使用模型来识别源代码和系统配置中的弱点。 事实证明，恶意行为者也可以这样做。

AI模式如何将网络安全转变为新工具，安全研究者使用OpenAI Group PBC的GPT-5.6 Sol扫描了PHP的代码基础，以识别错误。 他们发现了关键的脆弱性，各组织一直在为弥补这些脆弱性而奔波。

Amazon.com公司上周披露，攻击者在补丁发布之前四天开始针对PHP弱点，Cloudflare公司所看到的利用企图后来攀升到数百万。

最近一轮的数字威胁突出表明，各组织需要如何为不可避免的违反行为做好准备。 今年早些时候在拉斯维加斯举行的Dell Technology World聚会期间，有一个明确的信息，即数据保护和网络复原力已经成为企业AI的基础控制层。 根据CUBE研究公司Krista案，这为企业如何根据AI时代调整安全控制奠定了基础。

“这是将网络恢复能力与复原力、AI治理、数据完整性和业务信任联系起来的更广泛的市场对话的一部分，在这个世界上，企业AI系统与数据管道、运行时间控制和支持它们的恢复过程一样可靠，”Case 称。 “网络复原力和人工智能基础设施平台正在扩大，超越备份和恢复，进入元数据可见度、恢复验证和企业数据的背景理解。 各组织应密切注意供应商如何为AI环境建立治理和验证。”

这个特征是SiliconANGLE Media探索关键趋势和建筑变换动力连续，生产级AI的一部分。 确保检查AI网络安全和复原力峰会，10月27日推出由Dell Technologies牵头的旗舰方案，随后于10月28日推出包括InfoStale和Infinidat有限公司在内的更多行业视角。 这些广播将在10月28日至29日的CUBE、SiliconANGLE Media的直播工作室，(* 下文披露。 )

应对AI对网络安全的影响

在注重网络复原力的背后，网络安全界担心AI有可能在系统保护方面打开一个大洞。 AI模型大大提高了分析和潜在利用弱点的能力。 Microsoft公司的每月报告仅10月份就列出了1,453个新的弱点。

“AI改变整个游戏，” Steve Kenniston最近告诉CUBE, “这就是模式的诱惑。 有模式训练数据。 有一些系统使人们可以做提示词注入之类的事情。 需要考虑的是身份管理。

而戴尔则积极利用边疆模式建设防御工事。 该公司参与了Anthropic PBC的Glasswing项目，该项目使被审查的组织能够获得强大的AI工具来保障关键系统的安全。 Glasswing的目的是把软件支持关键基础设施的组织召集起来，如果成功受到攻击，可能导致灾难性后果。

AI agents的兴起也给企业系统中的认同带来了新的审查。 威胁行为者更容易获得AI工具，改变了准入管理和检测的接触规则。

戴尔的总裁兼首席保安官John Scimone最近与CUBE谈话说，“毫无疑问，入境费用和入境障碍已经破灭。 “称另一种方式是，明天最不尖端的攻击比一年前世界所见过的更尖端。”

数据复原力的重要性

在一个攻击日益尖端的世界中，另一个动态变得明显：没有警告。 企业面临的情况是，与飓风或海啸等自然灾害不同，全球许多地方都设有警报系统，而一场重大袭击往往不会预示未来会发生什么。

据产品管理信息系统主任Joseph D ' Angelo说，这更加强调恢复和复原的必要性。

称 D'Angelo最近与CUBE的谈话中，“因此，保护自己的唯一办法就是确保你拥有我们称之为可防的复原态势。 这就是了解你的应用状况，了解你的数据状况，能够以不为攻击所困或以任何方式损害的方式恢复这些应用。”

戴安格罗围绕数据的观点很重要。 目前为众多企业应用程序提供动力的AI系统，仅能与为其提供燃料的数据一样好。

数据需要存储，Infinidat等公司则专注于安全解决方案，随着AI提高网络风险的速度和复杂性，这些解决方案可以保护企业的业务状态。 这涉及不可改变的快照、网络检测和恢复能力相结合，旨在帮助各组织确定清洁恢复点，避免恢复受损数据。

称 Infinidat首席营销官Eric Herzog最近出现在CUBE时，“核心是数据，这就是为什么你需要在初级存储上进行下一代数据保护，这是AI将使用的方法。 在备份上也需要下一代数据保护。”

准备应对量子威胁

许多企业在网络安全方面的另一个关键发展涉及量子后加密，即PQC. 从古典加密向能够承受未来量子计算机攻击的量子安全算法的迁移，是任何应对新出现的威胁和加强长期网络复原力的战略的重要要求。

目前，企业似乎还没有准备好应对量子威胁。 今年早些时候由DigiCert Inc.发布的一项研究发现，被调查的组织中只有7%在其大多数数字证书中安装了量子安全或混合密码。 由于AI agents的成长大大提升了数据在企业平台中的作用，量子威胁最终会造成重大干扰。

“量子计算今天将打破保护数据的加密和数字签名，而代理AI 融资则通过增加数据的价值并自主地在团队和组织之间分享数据，从而打破其利害关系，”称 John Roese,Global CTO和Dell的AI首席官员。

AI工具为量子威胁的范围提供了洞察力。 7月下旬发表的Anthropic PBC报告详细介绍了使用Claude神话预览(英语：Claude Mythos Preview)的研究人员如何发现更好的攻击密码算法的方法。 其中包括成功削弱HAWK，这是国家标准和技术研究所的9名候选人之一，或者NIST的附加量子后数字签名算法程序。

称 Garrison Buss是QuSecure Inc.的首席技术官兼联合创始人，他在为SiliconANGLE准备的一次简报中警告说，“自2023年以来，CISA、NSA和NIST已经警告说，对手现在可以收集并持有加密数据，直到他们能够解密为止。 “人工智能结果的改变是时钟的另一面 — — 攻击者不再需要量子计算机来取得有意义的进步；他们只需要经典的计算和几天的时间。 ”

基础设施层的安全

随着2026年开始倒闭，企业面临尽快实施AI解决方案的需要，同时紧急防范人工智能带来的威胁。 这将需要基础设施层面的安全，从而能够减轻媒介AI、量子计算和复杂模型在威胁行为者手中构成的风险。

" 没有安全的基础设施，就不存在AI，这意味着数据也是可收回的，是安全的，当然也是符合要求的 "。 也开始出现在实际AI基础设施本身。 不要让备份和回收成为后脑勺

随着公司建立具有复原力的基础设施，未知因素可能是控制AI本身。 “流氓”AI agents对Hugging Face和联合国发动的大规模攻击说明了新自治世界的安全复杂性。 科技产业被数十年的创新所定义，这也许是网络安全方面非常规思维的时机。

“我们曾经画堆叠，安全是一层，”CUBE研究的Dave Vellante在最近一期CUBE Pod中表示，“这不再是一层。 它必须是正统的东西。 我认为攻击者总是有优势的。 你知道老话，他们只得说对一次也总会有缝合。

(* 披露：CUBE是AI网络安全和复原力峰会的付费媒体伙伴。) CUBE峰会报道的发起人对CUBE或SiliconANGLE上的内容没有编辑控制。 )

图片：SiliconANGLE/ChatGPT

SiliconANGLE联合创始人John Furrier的留言：

支持我们的使命，通过与CUBE社区接触，保持内容的开放和自由。 加入CUBE的校友信托网络，其中技术领导者相互连接，分享情报，创造机会。

- 15M+观看CUBE视频，为AI、云层、网络安全等各种对话提供动力

- 11.4k+CUBE校友——通过独特的信任网络与11400多名技术和企业领袖建立联系，塑造未来

你是AWS客户吗？在财务上支持SiliconANGLE，从我们的市场门户页面和链接购买您的AWS服务：https://siliconangle.com/aws-marketplace/.

关于 SiliconANGLE 媒体

SiliconANGLE Media由技术视觉家约翰·弗瑞尔(John Furrier)和戴夫·韦兰特(Dave Vellante)创立，构建了产业领先数字媒体品牌的动态生态系统，达到1500+百万精英技术专业人士。 利用CUBEAi.com神经网络，帮助科技公司以数据为导向做出决策

本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。

Cutting both ways: AI sharpens systems protection while opening new exploits for malicious actors

The cybersecurity industry is discovering that powerful frontier AI models are a double-edged sword.

Companies can use models to identify vulnerabilities in source code and system configurations faster than ever before. As it turns out, malicious actors can do this as well.

A prime example of how AI models have shifted cybersecurity into a new gear can be seen in recent news involving exploits surrounding the PHP programming language, which powers up to 70% of global websites. Security researchers used OpenAI Group PBC’s GPT-5.6 Sol to scan PHP’s code base for bugs. They discovered critical vulnerabilities, and organizations have been scrambling to patch them.

Meanwhile, Amazon.com Inc. disclosed last week that attackers began targeting PHP vulnerabilities four days before patches were released, and exploit attempts seen by Cloudflare Inc. have since climbed into the millions.

This latest round of digital threats highlights how organizations need to be prepared for the inevitable breach. During the Dell Technologies World gathering in Las Vegas earlier this year, there was a clear message that data protection and cyber resilience have emerged as foundational control layers for enterprise AI. This has set the stage for how enterprises must adapt security controls to the AI era, according to theCUBE Research’s Krista Case.

“This is part of a broader market conversation that ties cyber recovery capabilities to resilience, AI governance, data integrity, and operational trust, in a world where enterprise AI systems are only as reliable as the data pipelines, runtime controls, and recovery processes supporting them,” Case said. “Cyber resilience and AI infrastructure platforms are expanding beyond backup and recovery into metadata visibility, recovery validation and contextual understanding of enterprise data. Organizations should pay close attention to how vendors establish governance and validation for AI environments.”

This feature is part of SiliconANGLE Media’s exploration of key trends and architectural shifts powering continuous, production-grade AI. Be sure to check out the AI Cybersecurity and Resilience Summit, with a Dell Technologies-led flagship program Oct. 27, followed by additional industry perspectives, including InfoScale and Infinidat Ltd., on Oct. 28. The broadcasts will reach North America, EMEA and APJ audiences Oct. 28–29, on theCUBE, SiliconANGLE Media’s livestreaming studio. (* Disclosure below.)

Coping with AI impact on cybersecurity

Behind the focus on cyber resilience is a concern within the cybersecurity community that AI has the potential to open a significant hole in systems protection. AI models have dramatically increased the ability to analyze and potentially exploit weaknesses. A monthly report from Microsoft Corp. listed 1,453 new vulnerabilities in October alone.

“AI changes the whole game,” Steve Kenniston, senior cybersecurity evangelist for portfolio marketing at Dell, recently told theCUBE. “There’s the model inferencing. There’s the model training data. There are systems where people can do things like prompt injection. There’s identity management that needs to be thought about.”

For its part, Dell has been actively employing frontier models to build defenses. The firm is a participant in Anthropic PBC’s Project Glasswing, which gives vetted organizations access to powerful AI tools to secure critical systems. Glasswing’s purpose is to bring together organizations whose software supports critical infrastructure and could lead to catastrophic results if successfully attacked.

The rise of AI agents has also brought renewed scrutiny to identity in enterprise systems. Easier access to AI tools by threat actors has changed the rules of engagement for access management and detection.

“Undoubtedly the cost of entry and the barriers to entry have crumbled,” according to John Scimone, president and chief security officer of Dell, who spoke recently with theCUBE. “Said another way, the least sophisticated attack tomorrow is more sophisticated than the world will have ever seen a year ago.”

Importance of data resilience

In a world where the attacks have grown more sophisticated, another dynamic has become clear: there is no warning. Enterprises are confronted with a situation where, unlike natural disasters such as hurricanes or tsunamis, for which warning systems are in place in many parts of the globe, a major attack will often strike with no hint of what’s to come.

This has placed even greater emphasis on the need for recovery and resilience, according to InfoScale Director of Product Management Joseph D’Angelo.

“There’s no prediction here,” said D’Angelo, in a recent conversation with theCUBE. “So, the only way to protect yourself is to ensure that you have what we like to call a defensible recovery posture. And that is understanding the state of your applications, understanding the state of your data, and being able to recover those in a way that is not encumbered or in any way compromised by an attack.”

D’Angelo’s point around data is an important one. The AI systems currently powering numerous enterprise applications today are only as good as the data that fuels them.

Data needs storage, and companies such as Infinidat are focused on security solutions that can protect the operational state of a business as AI increases the speed and complexity of cyber risk. This involves a combination of immutable snapshots, cyber detection and recovery capabilities designed to help organizations identify clean recovery points and avoid restoring compromised data.

“It’s all about the data,” said Infinidat Chief Marketing Officer Eric Herzog, during a recent appearance on theCUBE. “The core is the data, which is why you need next-generation data protection on your primary storage, which is what AI would use. You need next-generation data protection on your backup, as well.”

Preparing for the quantum threat

Another key development on the cybersecurity horizon for many businesses involves post-quantum cryptography, or PQC. The migration from classical encryption to quantum-safe algorithms that can withstand attacks from future quantum computers is an important requirement for any strategy that addresses emerging threats and strengthens long-term cyber resilience.

At the moment, enterprises do not appear ready for the quantum threat. A study published earlier this year by DigiCert Inc. found that only 7% of organizations surveyed have deployed quantum-safe or hybrid cryptography across most of their digital certificates. As the growth of AI agents has significantly boosted the role of data in enterprise platforms, the quantum threat could ultimately cause significant disruption.

“Quantum computing will break the encryption and digital signatures protecting data today, while agentic AI raises the stakes by increasing the value of data and autonomously shares it across teams and organizations,” said John Roese, global CTO and chief AI officer at Dell.

AI tools have provided insight into the scope of the quantum threat. An Anthropic PBC report published in late July detailed how researchers using Claude Mythos Preview discovered improved ways to attack cryptographic algorithms. This included a successful weakening of HAWK, one of nine candidates in the National Institute of Standards and Technology, or NIST’s, process for additional post-quantum digital signature algorithms.

“CISA, NSA, and NIST have warned since 2023 that adversaries could be collecting encrypted data now and holding it until they can decrypt it,” said Garrison Buss, chief technology officer and co-founder of QuSecure Inc., in a briefing prepared for SiliconANGLE. “What the AI results change is the other side of that clock – an attacker no longer needs a quantum computer to make meaningful progress; they only need classical compute and a few days.”

Security in the infrastructure layer

As 2026 begins to wind down, enterprises are facing the need to implement AI solutions as fast as possible, while urgently protecting against the threats that artificial intelligence brings. This will require infrastructure-level security that can mitigate the risks posed by agentic AI, quantum computing and sophisticated models in the hands of threat actors.

“There is no AI without a secure infrastructure, and that means data that is also recoverable, that is secure, and is of course compliant,” said theCUBE Research’s Christophe Bertrand, during an analyst briefing. “Cyber resiliency risk is not just on the data, what you have to go through every day, it’s also starting to appear in the actual AI infrastructure itself. Don’t make backup and recovery an afterthought, because it’s going to hurt your wallet.”

As companies build resilient infrastructures, the unknown factor may be control over AI itself. Highly publicized attacks on Hugging Face and the United Nations by “rogue” AI agents illustrate the security complexities of the new autonomous world. The tech industry has been defined by decades of innovation, and this may be the time for unconventional thinking when it comes to cybersecurity.

“We used to draw the stack, and security was a layer,” said theCUBE Research’s Dave Vellante, during a recent edition of theCUBE Pod. “It’s not a layer anymore. It has to be something that’s orthogonal. I think that the attackers are always going to have an advantage. You know the old saying, they only have to be right once. And there’s always going to be seams.”

(* Disclosure: TheCUBE is a paid media partner for the AI Cybersecurity and Resilience Summit. Sponsors of theCUBE’s summit coverage do not have editorial control over content on theCUBE or SiliconANGLE.)

Image: SiliconANGLE/ChatGPT

A message from John Furrier, co-founder of SiliconANGLE:

Support our mission to keep content open and free by engaging with theCUBE community. Join theCUBE’s Alumni Trust Network, where technology leaders connect, share intelligence and create opportunities.

- 15M+ viewers of theCUBE videos, powering conversations across AI, cloud, cybersecurity and more

- 11.4k+ theCUBE alumni — Connect with more than 11,400 tech and business leaders shaping the future through a unique trusted-based network

Are you an AWS customer? Support SiliconANGLE financially by buying your AWS services from our Marketplace portal page and links: https://siliconangle.com/aws-marketplace/

About SiliconANGLE Media

Founded by tech visionaries John Furrier and Dave Vellante, SiliconANGLE Media has built a dynamic ecosystem of industry-leading digital media brands that reach 15+ million elite tech professionals. Our new proprietary theCUBE AI Video Cloud is breaking ground in audience interaction, leveraging theCUBEai.com neural network to help technology companies make data-driven decisions and stay at the forefront of industry conversations.

Auto-fetched and translated by AI Atlas. All rights belong to the original author.