Winning the information war is as much about ideas and perception as it is about data

A U.S. Marine with Joint Task Force-Ares introduces Ryan McBeth, an open-source intelligence analyst and cybersecurity expert, before a professional military education session at Fort George G. Meade, Maryland, July 23, 2026. (U.S. Marine Corps photo by Sgt. Isaiah W. Smith)

Information warfare is nothing new. What is new is the addition of social media and artificial intelligence (AI) making it possible to spread dis- and misinformation faster than ever, necessitating a rapid response to help maintain parity in the information battlespace.

Look no further than the Iran conflict, which has spawned a network of AI-generated videos claiming to be US servicemen and women. The University of Southern California’s Information Sciences Institute identified a network of more than 100 TikTok accounts impersonating US personnel, with more than 3 million views and 440,000 engagements on videos ranging from “soldiers” claiming to be distressed to asking for support for their mission.

There are numerous other forms of modern-day information warfare as well, from mocking and taunting such as Iran’s videos of US administration officials as Lego figures to propaganda in the Ukraine conflict. Being able to respond to a varied range of what has effectively become meme warfare necessitates quickly identifying target audiences, formulating the right message and determining the best way to get it in front of people, explained Bobby Muscara, associate director for Europe at Two Six Technologies, who is in the trenches with the company’s partners that depend on cutting-edge technologies to complete their missions.

Bobby Muscara, associate director for Europe, Two Six Technologies

“We have analysts who do routine threat analysis in the information environment, and what we’ve seen from both Russia and China is that they’re using AI and large language models (LLMs) to enhance their information operations campaigns,” Muscara said. “Whether it’s multiple languages – Russia, for example, was using two a few years ago, and now they’re up to six languages for the majority of their propaganda – and China is increasingly disseminating their propaganda in English and other languages. Other adversaries use AI as well. Iran, as we’ve seen recently with their ‘Lego’ videos. Our clients need to be able to react fast to those narratives the adversaries are pushing out and disseminate content as fast as possible.”

Winning on the information battlefield means being able to attack and counterattack quickly, with messaging that is compelling to the people who need to hear it. That isn’t as simple as dropping leaflets or radio broadcasts anymore. The tools have changed quickly, and understanding and using them effectively is critical to success.

The US military is embracing this challenge, as multiple branches have recently announced new initiatives or changes in how they conduct information warfare. In late 2025, the Navy launched a pilot program to incorporate information warfare squadrons within carrier groups and the Army announced it was adding electronic warfare and information advantage to the responsibilities of its principal cyber advisor.

In July, the Marine Corps announced that it is working on an information warfare plan that will encompass the entire Corps, building a framework that will encompass training, funding requests and sharing information with industry representatives to identify the information warfare tools the Corps needs.

“There is an insatiable demand and realization by combatant commanders now that non-kinetic effects are integral to what they need to do, and they are turning to the services, particularly to the Marine Corps, to say, ‘HI need this capability in addition to the traditional arms of the’” Marine Air-Ground Task Forces, Deputy Commandant for Information Lt. Gen. Joseph Matos said at the Modern Day Marine Conference in April. “Those demand signals are out there because they realize, in order for the combat commander to do his job, they need these non-kinetic effects from across the services.”

AI tools are providing new information warfare capabilities that demand immediate, trustworthy responses. Deepfakes and misinformation campaigns aim to not only control the narrative on the battlefield, but destabilize governments or sway public opinion. While fighting on the battlefield may be contained to one region, the narrative surrounding it has no geographic constraints.

“Frontier AI capabilities can be used by potential adversaries to enhance cyber operations, generate convincing disinformation at scale, or support targeting in ways that compress decision timelines and increase escalation risks,” Adm. Richard Correll, Commander, STRATCOM said at the Defense Intelligence Agency’s 2026 DoDIIS Conference. “The question is not whether AI will be present in the future conflict. It already is.”

When a campaign is gaining traction and needs to be countered, a rapid response is critical to effective counter-messaging. Muscara points to three tools: understand the information environment, create the content needed to counter a campaign and disseminate it. Two Six’s Sentr – the company’s recently launched agentic AI-driven “command center” for the narrative battlefield – combines all three of those capabilities into one platform, streamlining the process for identifying propaganda, drafting a counter and distributing it effectively.

Using both agentic AI and LLMs, Sentr can scour social media sites and the larger internet to compile data on active campaigns infinitely faster than humans, compile metrics on the effectiveness of those campaigns and track how they are evolving.

Once a counter message is decided on, Sentr can distribute it and track its effectiveness, as well, giving users the ability to modify the message as needed to ensure it is resonating with the target audience. With one platform offering all of the necessary capabilities, Muscara said customers have seen response time cut by 85 percent with Sentr.

“The crucial thing is that we can inject information into the information environment all with one tool, rapidly, without needing to coordinate between three different types of tools or vendors to get after one problem,” he said. “That rapid lifecycle is done in-house in one platform, and that’s how we can enable the customer to rapidly disseminate content.”

U.S. Army Lt. Col. Jason Welch, United States Indo-Pacific Command, asks a question during the Artificial Intelligence (A.I.) systems for Public Affairs brief, at the Public Affairs Leadership Development Forum (PALDF), in the Hyatt Regency Waikiki Resort Conference Room, Mar. 31, 2026. The A.I. systems brief educated attendees on the current systems programs and resources available to Public Affairs professionals across the theater. (U.S. Army Photo by Sgt. Qishaunia Hawkins)

Identifying the narrative

With so much information flying around on a daily basis for everyone – regardless of whether they are in a contested or potentially contested area or not – it is hard to separate what is trolling or people making memes for fun versus potentially serious disinformation or misinformation campaigns. Social media apps enable instantaneous sharing of information, both true,false or twisted, but when false information is corrected, the correction often receives much less attention than the original post.

Or, as the old saying goes – often misattributed to Mark Twain – “a lie gets halfway around the world before the truth gets its shoes on.”

The first step to combating the information flow is identifying the core narrative. With so much information being shared so quickly, not everything can be fought against. Sentr gets to the root of the information campaign and compiles the necessary data in a way that it can be viewed holistically, said Muscara.

“Once we collect that small data lake of curated data that matters to our customers, the AI can rapidly identify those narratives by asking simple prompts such as ‘what narratives is Russia pushing in Lithuania over the last seven days?’ and it’ll bring back language-coded responses showing you what Russia has pushed from all different platforms, as well as citations, so you can go back and look at the original content,” said Muscara. “That’s a way for our customers to rapidly assess the information environment as opposed to going to various websites or social media platforms. It’s all in one place.”

When that information is collected, Sentr’s narrative tools identify common patterns, such as wording, hashtags and target audiences. Then, users can craft a counter-narrative to hit those common areas, allowing for an effective messaging campaign to combat the propaganda.

In some cases, the best path of action is inaction. If a narrative isn’t gaining traction, it can be monitored without a counter-messaging campaign that could draw attention to a false narrative. By tracking the volume of the messaging, engagement and popularity, Sentr helps users maximize their bandwidth – both literally and figuratively – by only developing campaigns when necessary.

“Being able to tell the commander you don’t need to do something is actually a pretty powerful message, as well, and our platform can do that,” said Muscara.

Security and reliability

Sentr is a powerful tool for building and implementing campaigns, which means it is potentially dangerous if it can be hacked or otherwise interfered with. Cybersecurity is a primary concern for the company and its users, and beyond that, the potential for AI hallucinations are also top of mind.

Any AI tool is only as good as the information it draws from. Sentr is not working from an off-the-shelf LLM that could have potentially false data. Instead, it uses what the company calls a full-doctrinal backbone, where each job is carried out under defined doctrinal parameters such as where the information is coming from and how it is supported that set standards for information should be used and how it should be used. The platform also uses a citations tool that provides context and supporting information for the reporting it does on campaigns, their key components and their reach.

The citations tool offers two methods of verification for users to understand where information is coming from. It has a curated knowledge base that shows users which document is being cited and what information is being pulled from that document. It also can cite articles from the internet, in which case it shows the citation and provides links, accompanying images, authors, the original text and any translated text.

“We can actually show you the real content,” said Muscara. “It’s not just citing a law book in its database. It’s showing real-time or near real-time information of the information environment – and the AI can’t hallucinate a web page.”

All of Sentr’s information flows are curated by human experts, who identify relevant information and build relevant context around it to ensure that users are getting the best possible information to build campaigns. Sentr also tracks history and allows users to audit where information originated as users build out their documentation. Every piece of content can be viewed from the same interface, giving a broad perspective on where information came from, how it was collected and supporting citations to help validate it. The platform is designed to include a human in the loop, allowing for independent evaluation and verification of what Sentr is reporting.

Fundamental cybersecurity is part of the platform, as well. Sentr is an unclassified tool, but the company uses the same security protocols as it does for its systems that operate on classified networks. Two-factor authentication is required, and Muscara pointed out that many of the company’s customers are also well-trained in tradecraft for things such as protecting login info that they apply when they are using the platform.

Sentr can also identify potential leaks or cybersecurity issues when it is compiling data. For example, if the platform is searching for a narrative and identifies related information coming from a friendly source, it can flag it as an operation security (OPSEC) issue, said Muscara.

“You can find it by using the broadbased LLM that’s searching the data lake for this news article or tweet or whatever it was,” said Muscara. “Then you can pivot to create an OPSEC plan to mitigate this or suggest additional training or TTPs that we should implement and send out to all soldiers within the unit.”

Winning the information war

Fighting misinformation and disinformation – or even just correcting information that was thought to be true and later turned out to be incorrect or not quite accurate – is a constant challenge. Whether it’s memes that appear harmless on the surface but are subtly pushing a narrative or full-on campaigns aimed at swaying public opinion or destabilization, adversaries are harnessing AI tools to quickly pump out staggering volumes of information that must be analyzed and countered to avoid ceding control of the narrative.

Doing so requires tools that can offer information specialists the ability to identify and counter emerging narratives before they can take root. By providing a single platform that can encompass all aspects of identifying and responding to narratives, Two Six’s Sentr enables human experts to combine their knowledge with AI tools to respond quickly.

“Our message to customers and potential customers is this is an enabler for who you already have and the experts that you already have,” said Muscara. “We’re making our partners faster, more reliable and more impactful within the platform and the information environment.”

2026年7月23日，在马里兰州乔治·米德堡举行的专业军事教育课程开始前，一名隶属于联合特遣部队-阿雷斯的美国海军陆战队员介绍了开源情报分析师兼网络安全专家瑞安·麦克贝斯。（美国海军陆战队中士以赛亚·W·史密斯摄）

信息战并非新鲜事物。真正的新变化在于社交媒体和人工智能（AI）的加入，使得虚假信息和错误信息的传播速度前所未有，因此需要迅速做出反应，以维持信息战的平衡。

伊朗冲突就是一个典型的例子，这场冲突催生了一个人工智能生成的视频网络，这些视频冒充美国军人。南加州大学信息科学研究所发现，有超过100个TikTok账号冒充美国人员，这些视频的观看量超过300万次，互动量超过44万次，内容从“士兵”声称自己处境艰难到请求支援，无所不包。

现代信息战的形式多种多样，从伊朗将美国政府官员比作乐高人偶的嘲讽视频，到乌克兰冲突中的宣传，不一而足。Two Six Technologies 欧洲区副总监 Bobby Muscara 解释说，要应对这种实际上已经演变为“迷因战”的各种攻击，就必须迅速识别目标受众，制定正确的信息，并找到最佳的传播方式。Muscara 与公司的合作伙伴并肩作战，这些合作伙伴依靠尖端技术来完成任务。

Bobby Muscara，Two Six Technologies 欧洲区副总监

“我们有分析师负责对信息环境进行常规威胁分析，我们发现俄罗斯和中国都在利用人工智能和大型语言模型（LLM）来增强其信息战行动，”穆斯卡拉说道。“无论是多种语言——例如，俄罗斯几年前只使用两种语言，而现在他们的大部分宣传都使用六种语言——还是中国越来越多地使用英语和其他语言传播宣传。其他对手也在使用人工智能，例如我们最近看到的伊朗的‘乐高’视频。我们的客户需要能够快速应对对手散布的这些信息，并尽快传播相关内容。”

赢得信息战的关键在于能够快速发起进攻和反击，并向目标受众传递具有说服力的信息。这不再像以前那样简单地散发传单或进行广播宣传了。信息传播工具日新月异，理解并有效运用这些工具是成功的关键。

美国军方正积极应对这一挑战，多个军种近期都宣布了信息战方面的新举措或变革。2025年底，海军启动了一项试点项目，将信息战中队纳入航母战斗群；陆军也宣布将电子战和信息优势纳入其首席网络顾问的职责范围。

7 月，美国海军陆战队宣布正在制定一项信息战计划，该计划将涵盖整个海军陆战队，建立一个框架，其中包括培训、资金申请以及与行业代表共享信息，以确定海军陆战队所需的信息战工具。

“作战指挥官们现在迫切需要并意识到非动能作战手段是他们作战行动不可或缺的一部分，他们正向各军种，特别是海军陆战队，提出‘除了传统的武器装备外，海军陆战队空地特遣部队也需要这种能力’，”海军陆战队信息副司令约瑟夫·马托斯中将在四月份的现代海军陆战队会议上表示。“这些需求信号之所以出现，是因为他们意识到，为了让作战指挥官履行职责，他们需要各军种提供这些非动能作战手段。”

人工智能工具正在提供全新的信息战能力，这就要求我们做出迅速且可信的回应。深度伪造技术和虚假信息宣传活动的目标不仅在于控制战场上的舆论，还在于破坏政府稳定或左右公众舆论。虽然战场上的战斗可能局限于某一特定区域，但围绕战斗的舆论却不受地域限制。

“潜在对手可以利用前沿人工智能能力来增强网络作战能力，大规模制造逼真的虚假信息，或以缩短决策时间、增加冲突升级风险的方式支持目标定位，”美国战略司令部司令理查德·科雷尔海军上将在国防情报局2026年国防部信息系统与战略会议（DoDIIS）上表示。“问题不在于人工智能是否会在未来的冲突中出现，它已经出现了。”

当一场宣传活动势头强劲，需要进行反制时，快速响应对于有效的反制信息至关重要。穆斯卡拉指出了三种工具：了解信息环境、创建反制宣传活动所需的内容以及传播这些内容。Two Six 公司最近推出的 Sentr——该公司基于人工智能的“指挥中心”，专为叙事战场打造——将这三种功能整合到一个平台中，从而简化了识别宣传、撰写反制内容以及有效传播的流程。

Sentr 利用智能 AI 和 LLM，可以比人类快得多地搜索社交媒体网站和更广泛的互联网，收集有关活跃活动的数据，收集有关这些活动有效性的指标，并跟踪它们的发展情况。

一旦确定了反制信息，Sentr 即可分发并追踪其效果，用户还可以根据需要修改信息，确保其能引起目标受众的共鸣。Muscara 表示，Sentr 平台提供所有必要功能，客户的响应时间已缩短了 85%。

他说：“关键在于，我们只需一个工具就能快速地将信息注入信息环境，无需协调三种不同的工具或供应商来解决同一个问题。这种快速的生命周期管理在一个内部平台上完成，这正是我们能够帮助客户快速传播内容的方式。”

2026年3月31日，在威基基凯悦酒店会议室举行的公共事务领导力发展论坛（PALDF）上，美国印太司令部陆军中校杰森·韦尔奇在人工智能（AI）系统公共事务简报会上提问。此次人工智能系统简报会旨在向与会者介绍当前可供战区公共事务专业人员使用的系统程序和资源。（美国陆军照片，摄影：Qishaunia Hawkins中士）

确定叙事方式

如今，无论身处冲突地区还是潜在冲突地区，每天都有大量信息涌入，难以分辨哪些是网络喷子或人们为了娱乐而制作的表情包，哪些是可能造成严重后果的虚假信息或误导性信息传播活动。社交媒体应用使得信息的即时分享成为可能，无论真假或被歪曲，但当错误信息被纠正时，纠正内容往往远不如原帖受到关注。

或者，正如那句常被误认为是马克·吐温说的古老谚语所说：“谎言跑遍半个世界，真相才穿好鞋子。”

对抗信息流的第一步是识别核心叙事。由于信息传播速度极快，无法对所有信息进行有效应对。Sentr 致力于挖掘信息传播的根源，并以整体视角整理必要数据，Muscara 说道。

“一旦我们收集到对客户至关重要的精选数据，人工智能就能通过简单的提问，例如‘过去七天俄罗斯在立陶宛散布了哪些信息？’，快速识别出这些叙事。它会返回带有语言编码的答案，展示俄罗斯在各个平台上发布的信息，以及相关的引用，方便用户查看原始内容，”穆斯卡拉说道。“这样一来，我们的客户无需访问各种网站或社交媒体平台，就能快速评估信息环境。所有信息都集中在一个地方。”

收集到这些信息后，Sentr 的叙事工具会识别出常见的模式，例如措辞、话题标签和目标受众。然后，用户可以针对这些共同点构建反叙事，从而开展有效的宣传活动来对抗宣传攻势。

在某些情况下，最好的做法是不采取行动。如果某种说法没有获得关注，可以对其进行监控，而无需发起反制宣传活动，以免将注意力引向错误的说法。Sentr 通过追踪信息量、互动情况和受欢迎程度，帮助用户最大限度地利用资源——无论是字面意义上的还是比喻意义上的——仅在必要时才开展宣传活动。

“能够告诉指挥官你不需要做某事，这实际上也是一个非常有力的信息，而我们的平台可以做到这一点，”穆斯卡拉说。

安全性和可靠性

Sentr是一款功能强大的营销活动构建和执行工具，这意味着如果它遭到黑客攻击或以其他方式干扰，就可能造成危险。网络安全是该公司及其用户最为关注的问题，除此之外，人工智能可能引发的幻觉也是他们最为担忧的。

任何人工智能工具的效用都取决于其所依赖的信息质量。Sentr 并非使用现成的、可能包含错误数据的逻辑模型（LLM），而是采用该公司所谓的“完整理论框架”。在这个框架下，每项任务都严格遵循既定的理论参数，例如信息的来源、支持方式以及信息的使用标准。此外，该平台还配备了引文工具，为其关于竞选活动、关键组成部分及其影响范围的报告提供背景信息和佐证资料。

该引文工具提供两种验证方法，帮助用户了解信息的来源。它拥有一个精心整理的知识库，可以显示引用的文档以及从中提取的信息。此外，它还可以引用互联网上的文章，并显示引文信息，包括链接、相关图片、作者​​、原文以及任何翻译文本。

“我们实际上可以向您展示真实的内容，”穆斯卡拉说。“它不仅仅是引用数据库中的法律书籍。它展示的是信息环境的实时或近实时信息——人工智能无法凭空捏造网页。”

Sentr 的所有信息流均由专家人工审核，他们负责识别相关信息并构建相关背景，以确保用户获得最佳信息以开展营销活动。Sentr 还会追踪信息历史记录，并允许用户在完善文档的过程中审核信息的来源。所有内容均可通过同一界面查看，从而全面了解信息的来源、收集方式以及有助于验证信息的引用依据。该平台的设计理念是引入人工参与，以便对 Sentr 报告的内容进行独立评估和验证。

基础网络安全也是该平台的重要组成部分。Sentr 是一款非机密工具，但该公司采用的安全协议与运行在机密网络上的系统相同。系统要求启用双因素身份验证，Muscara 指出，该公司许多客户都接受过专业的安全培训，例如在使用该平台时如何保护登录信息。

穆斯卡拉表示，Sentr在收集数据时还能识别潜在的泄露或网络安全问题。例如，如果平台在搜索某个叙事线索时发现来自友方来源的相关信息，它就可以将其标记为行动安全（OPSEC）问题。

“你可以利用覆盖面广的LLM（数据湖管理工具）搜索数据湖，找到这条新闻报道、推文或其他任何内容，”穆斯卡拉说。“然后你可以迅速制定行动安全计划来缓解这种情况，或者建议我们应该实施并向部队所有士兵发送额外的培训或战术、技术和程序。”

赢得信息战

打击虚假信息和错误信息——甚至仅仅是纠正那些一度被认为是真实但后来被证明是错误或不准确的信息——始终是一项挑战。无论是表面上看似无害却暗含某种叙事意图的表情包，还是旨在左右舆论或破坏社会稳定的全面宣传活动，对手都在利用人工智能工具迅速炮制海量信息，而这些信息必须经过分析和反驳，才能避免失去舆论控制权。

要做到这一点，需要一些工具，使信息专家能够在新出现的叙事扎根之前识别并应对它们。Two Six 的 Sentr 提供了一个涵盖识别和应对叙事所有方面的单一平台，使人类专家能够将他们的知识与人工智能工具相结合，从而快速做出反应。

“我们想向客户和潜在客户传达的信息是，这能更好地发挥你们现有的资源和专家的作用，”穆斯卡拉说。“我们让合作伙伴在平台和信息环境中更快、更可靠、更有影响力。”