The escalating dispute between Anthropic and the United States Department of War is becoming more than a government procurement fight. For advertisers and tech buyers, it’s an early signal that the AI platforms powering future media and commerce may increasingly be shaped by geopolitical alliances.

On Monday, Anthropic sued the Pentagon after the department designated the company a “supply chain risk,” a label typically reserved for foreign adversaries. The designation requires companies doing work tied directly to the department to stop using Anthropic’s flagship product, Claude.

The move followed tensions between the two after the government previously used Claude in operations, including the ongoing war in Iran. Anthropic argues the designation punishes the company for its positions on AI policy and exceeds the government’s authority.

The case is part of a broader clash playing out as AI companies deepen ties with Washington while competing fiercely for commercial users. Meanwhile, OpenAI—whose AI answer engine ChatGPT dominates consumer adoption, reaching 900 million weekly active users—has expanded its own relationship with the U.S. government after accepting a deal Anthropic previously declined.

The moment underscores how AI firms are increasingly navigating two powerful constituencies: governments seeking strategic AI partnerships and a global consumer base that expects neutrality from the tools they use every day. 

For brands, the standoff adds a new layer of uncertainty around brand safety, as AI infrastructure becomes an increasingly critical channel for reaching consumers online.

AI’s Digital Nation-States

For an ad industry that is increasingly reliant on AI products rolled out by these companies, the implications may stretch far beyond defense contracts.

According to Nicole Greene, vice president, analyst at Gartner, governments and AI companies are increasingly operating as strategic blocs—what she calls “digital nation-states.”

Those alliances blur the line between technology infrastructure and geopolitics.

Large tech vendors are spending more than $70 billion per quarter on AI infrastructure, according to Gartner’s analysis, with six major tech companies, including Amazon, Google, Microsoft, Alibaba and Oracle, investing over $300 billion in 2025. That scale rivals the economic output of many countries, Greene pointed out.

For advertisers and enterprises that rely on AI platforms, Greene said, that means scrutinizing not just product capabilities but also the political and strategic relationships behind them.

“If you’re a U.S company, are you going to be comfortable using a model that has a strategic alliance with China?” Greene said. “These are global infrastructures that are going to have these new partnerships with governments that’s going to impact how much customers, not only trust that platform, but how much they trust the media being distributed through that platform.”

In that sense, the Anthropic dispute is less about one contract and more about how AI platforms will operate within evolving geopolitical ecosystems.

Brand Safety Risks

The Pentagon’s designation has already led to brands in the regulated sector—such as banks and financial institutions—evaluting its compliance reviews with Anthropic. 

“Anybody who is close to government services has had to suddenly assemble a war room like an emergency session of subject matter experts and decision makers and evaluate the risk,” said Brian Bauer, vice president of AI products at Rational Exponent, which works with banks and financial institutions. 

For advertisers, the standoff also underscores that AI models increasingly carry brand identities tied to their creators. Models like Claude and ChatGPT often reflect the positioning of their parent companies—whether emphasizing safety, openness or rapid innovation. That identity shapes how users and businesses perceive them.

ChatGPT uninstalls in the U.S surged 295% day-over-day on February 28, as people responded to the news of the AI company’s deal with the Department of War, according to Sensor Tower. 

Bauer said sentiment toward AI models typically develops over time and is unlikely to change overnight because of a single policy dispute. Still, controversies involving governments or national security could shift perception if a major incident occurs.

“A material event in the future—a misstep by the military where something unexpected and less than positive happens and it’s traced back to a root cause that could be associated with one of these vendor AI models,” Bauer said. “At that point, people might say it’s really hard for me to be associated with this product.”

According to CBS, the U.S was likely responsible for a strike that hit a girls’ school in Iran, killing 168 people, many of them children.

Why Advertisers Should Be Watching

For brands exploring advertising on ChatGPT or commerce within AI answer engines, analysts said the episode is a reminder that the environment surrounding those tools is as important as the technology itself.

Jacob Bourne, technology analyst at Emarketer, said “The real issue is political and reputational optics. Brands that are advertising and looking to advertise on ChatGPT, for example, should be aware that this is a consumer sentiment issue, ultimately.”

The dispute also surfaces long-standing questions around AI governance—specifically whether companies or governments ultimately set the rules for how these systems are deployed. For advertisers, the answer may shape where the next generation of digital media platforms emerges—and who controls them.

“Its bigger than advertising– it’s about platform stability, governance, and brand safety” Greene said. “For advertisers its an early signal on how these AI platforms will evolve over the political, regulatory environment and ethical risk conversation.”

Stagwell expects first quarter 2026 new business to be the strongest in its history, giving the holding company reason to be optimistic about the year ahead, according to CEO Mark Penn.

“I can say with some confidence that our Q1 2026 net new business is shaping up to be the strongest in the history of the firm,” Penn said during Stagwell’s fourth quarter and full year earnings call Tuesday.

Penn attributed the opportunity to upheaval at larger rivals. “We see great opportunity in 2026 to capitalize on an industry distracted by restructurings and mergers, and bolster our position as a winner in the age of AI,” he said.

The remarks came alongside Stagwell’s latest financial results, which showed modest revenue growth and stronger profitability.

The Numbers

• $807 million: Q4 2025 revenue (+2% year over year)
• $651 million: Q4 net revenue (+3%)
• $129 million: Q4 adjusted EBITDA; ~20% margin on net revenue
• $0.30: Q4 adjusted diluted EPS
• $2.9 billion: Full-year 2025 revenue (+2%)
• $2.43 billion: Full-year net revenue (+6%)
• $422 million: Full-year adjusted EBITDA; 17% margin on net revenue
• $0.83: Full-year adjusted diluted EPS

Watercooler Talk

Stagwell’s 2025 results reflect steady growth across its core businesses as the company pushes further into AI-powered marketing tools.

Full-year net revenue rose 6% to $2.43 billion, while adjusted EBITDA reached $422 million at a 17% margin. Net income attributable to common shareholders jumped to $29 million from $2 million the prior year.

Growth was led by digital transformation (+13% net revenue) and marketing services (+6%). Stagwell’s Marketing Cloud unit had a standout year, with net revenue surging 230%.

CFO Ryan Greene credited improved efficiency and cost discipline for the margin gains. “2025 marked an inflection year for Stagwell, with clear momentum in the underlying business and improving efficiency contributing to strong year-over-year net revenue, adjusted EBITDA, and adjusted EPS growth,” he said.

The company also expanded its stock repurchase program by $350 million, bringing total authorization to $725 million with about $400 million remaining.

For 2026, Stagwell is projecting net revenue growth of 8% to 12% and adjusted EBITDA of $475 million to $525 million.

Key Quote

“With industry consolidation and chaos, we’re seeing increased opportunities to win new larger wins,” Penn said.

When a founder wires $150 million of his own money into a stock that is down 60 to 70 percent, he is not just betting on his company. He is betting on how media will be bought in the future. 

Jeff Green’s open market purchase of roughly 6 million shares of The Trade Desk, about $148 to $150 dollars in personal capital, comes after a brutal derating across adtech and high-multiple software. 

The Trade Desk’s revenue is still growing in the teens, but it has clearly slowed, and 2026 guidance landed softer than investors wanted. Add in noisy debates around pricing and competition, plus a market still trying to price what large language models like Claude and ChatGPT mean for incumbent platforms, and the instinctive response has been simple: mark down the stock.

Because, you do not pay peak era valuations for what Wall Street views as yesterday’s model.

Green is betting that The Trade Desk becomes infrastructure

Green is being more contrarian about what comes next. Across the industry, early AI systems are already taking over work that used to require teams, interpreting briefs, ingesting performance data, reallocating budgets, and tweaking campaigns across CTV, open web, and retail media. These AI systems often have faster planning cycles and tighter feedback loops than human-only workflows. 

As these automated agents scale, they will still need neutral platforms they can trust for identity, access to quality supply, and reliable performance data.

That is where independent infrastructure like The Trade Desk matters. 

It looks less like Green is wagering on a simple rebound in software, and more like a shift toward automated, agent-driven media buying in which The Trade Desk becomes one of the core rails those systems run on. 

Green is buying in at a time when the market sees slower growth, tougher competition, and macro uncertainty, because he is wagering that the street is underestimating The Trade Desk’s role in that automated future.

That context makes The Trade Desk’s strategy more interesting. 

Green has argued that AI does not replace independent platforms, it makes them more powerful if they are open, neutral, and wired into enough data and supply. 

Despite their limited success during their initial rollouts, initiatives like UID2, OpenPath, and the recently unveiled OpenTTD point in the same direction. The company is increasingly positioning itself as infrastructure for identity, measurement, and access to premium inventory, exposed through APIs that other tools and agents can build on. 

We have seen signs that The Trade Desk aims to acquire the decisioning layer above the bidstream—advanced planning and modeling capabilities that inform advertiser strategies and influence automated execution. 

3 reasons some adtech firms will endure despite AI

While general purpose AI will make some software companies easy to replace, it’ll be hard to replace adtech companies that offer three things: unique demand and conversion data; closed loop measurement over years of campaigns/ and the plumbing into CTV, open internet, and retail media. 

If The Trade Desk can continue to strengthen that combination and add more decisioning capabilities, it will look less like a point solution, and more like core infrastructure for an automated buying stack. 

So, Green’s $150 million trade is not a bet that The Trade Desk magically reclaims its peak 2021 multiple. 

Even for the winners in this cohort, that is not the right base case. 

It is a bet that as advertising becomes more automated, a handful of well-connected, data rich platforms will continue to show value, because they’ll be the pipes that conduct money across the open web and CTV.

In the very short term, Green’s trade has already worked. 

Between the disclosure of the buy and renewed optimism around AI-driven inventory, the stock move has created about tens of millions of dollars in paper gains for Green. 

That’s more a byproduct than the point. 

The core of his move is to signal that, despite real risks and a skeptical market, he believes investors aren’t giving enough credit to the value of automated open web media buying, and that The Trade Desk’s current valuation does not reflect its chance to win in that future.

Investors can disagree with that view. What they cannot say is that he failed to put a clear price on it.

Brands and agencies have no shortage of AI tools for creative tasks—what they lack is cohesion. Creative teams often jump between image generators, video tools and language models, stitching outputs together manually and coordinating across departments. That patchwork process is what Luma AI says it wants to fix.

The Palo Alto–based company, founded in 2021, is launching a new category of end-to-end “creative agents” designed to coordinate and execute professional creative work across text, image, video and audio.

Luma spent its first three years building native 3D, image and video models. Now, the company is developing unified systems that work across modalities, with a specific focus on creative output for brands and agencies. The move signals Luma’s ambition to become a core part of the workflows for enterprise creative teams. 

“A couple of years back, most foundational model companies made the bet that the way toward intelligence was large language models,” said Caroline Ingeborn, COO of Luma AI. “While that’s true—with coding and deep research—that’s a sliver.”

The company has beta tested the agent with more than 100 clients since December. Launch partners include Serviceplan Group and Publicis Groupe Middle East & Turkey. 

From linear to agent-led

Until now, Ingeborn said, most creative teams have used AI tools for isolated parts of the creative process—generating an image, drafting copy or editing a video. She described the process as a “fragmented” approach.

The new agents are designed to collaborate across the entire workflow.

The system integrates Luma’s proprietary image and video models alongside third-party systems such as Veo, DeepSeek, ElevenLabs and Kling, with the ability for teams to toggle specific models on or off. The goal, Ingeborn said, is to eliminate the constant switching between tools.

Today’s creative production, she argued, remains highly sequential.

“It’s upending this process that has been very linear and very expensive, from ‘I have a script, then I do a mood board, then I do a video,’” Ingeborn said. “And for every one of these steps your project either gets killed, or you get more money.”

Luma, which has raised a total of $1.1 billion at a $4 billion valuation, positions itself as building toward multimodal general intelligence. Its latest offering arrives as more companies push into creative AI infrastructure, including Higgsfield, founded by Snap’s former generative AI chief.

However, Ingeborn argued Luma’s differentiation lies in its end-to-end execution layer.

“We are seeing that the process is moving from being very linear to being non linear,” said Igenborn. “The mission is that we want to build intelligence that can generate and operate and create alongside us humans.”

Ad agencies, like Havas, Broadhead and Supergood, are vibe-coding their own generative engine optimization (GEO) tools on top of large language models—often in a matter of hours.

Using coding assistants like Anthropic’s Claude Code, teams are building bespoke applications that analyze how brands appear in AI-generated responses, track competitors and, in some cases, package those tools into products sold to clients.

One example is Havas’s Brand Insights AI, a GEO product built using Claude Code and Replit. The tool generates prompts based on a client’s brand, runs them across multiple models, and analyzes how often a brand appears in responses, including citations—effectively simulating how a brand shows up in AI-driven discovery.

The platform has been rolled out globally, covering nearly 100 countries and more than 60 languages, and is licensed to clients as a SaaS product. It has also become a core part of the agency’s pitch strategy and helped win new business, according to Dan Hagen, Havas’ global chief data and technology officer.

The push to build GEO tools comes as brands try to influence how they appear in AI-generated answers, with more people turning to platforms like ChatGPT to find information. That shift has sparked a wave of startups—including Profound, Bluefish and Emberos—promising to help brands track and improve their visibility in AI responses. But three agencies interviewed for this story said they are increasingly building their own systems, arguing that off-the-shelf tools don’t fit how their teams work.

For Hagen, the appeal of building in-house comes down to control. Rather than adapting to third-party platforms, the agency can tailor features for specific use cases from brands managing multiple portfolios to teams in SEO or PR.

“You have so much control over the interface and the way you can build against it,” he said. 

Havas has so far opted against signing an exclusive enterprise agreement with Anthropic, which Hagen said can run into “multiple millions” annually. “It’s a combination of flexibility. It would be challenging for me to sign four or five enterprise agreements just from weight of cost,” he said, noting that pricing structures often fluctuate based on usage volumes, token consumption and model type.

Hagen also pointed to “cost control and management,” given the uneven adoption across the agency. While some employees are deeply embedded in AI workflows, others are still early in the learning curve. Committing to thousands of enterprise licenses, he said, risks paying for capacity that isn’t yet fully utilized. “We didn’t want to be in a position where we’re paying for ten thousand licenses that people are using once a week,” he said. “We didn’t want to get ourselves in a position where we’re sort of bedded in with one enterprise level deal with whoever and then that becomes a solution. What if they’re not frontier enough in six months–that puts us in a difficult situation.”

Creating an application within hours

The independent agency Broadhead is also experimenting with vibe coding GEO tools.

VP of product innovation Mitch Hislop said he “vibe coded” the first version of the agency’s GEO monitoring platform in a single evening using Claude Code. The tool analyzes how different AI providers rank a brand and its competitors.

One of its earliest features was what the team calls a “competitive intelligence vote,” where a user inputs a brand and location, and an LLM returns the competitors it is most likely to surface. The team then extended the feature by layering in audience personas—allowing the system to simulate how different types of consumers might query tools like ChatGPT or Claude, and how each brand ranks in those responses.

That upgrade took about two hours, Hislop said.

The result is a more dynamic form of competitive analysis, showing not just who a brand competes with, but how that competitive set shifts depending on user intent.

For Hislop, the advantage of building in-house is flexibility. “We could use what SEMrush provides, but we don’t like A, B and C about it. We don’t want to pay for SEMrush and Profound,” he said. “Instead, we have our own solution. We can make it work exactly how we want.”

Turning models into infrastructure

Other agencies are going further, adopting Anthropic’s models as part of their core infrastructure.

Mike Barrett, founder and chief strategy officer at Supergood, said the agency has an enterprise agreement with Anthropic and uses its models via API across a range of applications, including organizing internal knowledge graphs and shaping how brands show up in AI-generated search results.

In this setup, a model generates a response, evaluates it against predefined criteria, assigns a score and repeats the process until it reaches a target threshold—effectively acting as both creator and editor.

The process allows models to improve outputs over multiple passes without human intervention, Barrett said.

“Everybody’s making software right now,” Barret said. “In two years we are going to be delivering more software than actual documents.

Havas Media Network North America has appointed Sharona Sankar-King as chief data and product officer, the agency told ADWEEK.

Sankar-King joins from marketing services firm Harte Hanks, where she served as chief customer and data officer, and will report directly to North America CEO Greg James. The hire gives Sankar-King oversight of Converged.AI, the agency’s proprietary AI platform, and its broader analytics practice across the U.S. and Canada.

Sankar-King’s appointment comes as Havas accelerates investment in AI infrastructure and positions Converged.AI as a differentiator in a market increasingly defined by data complexity and privacy fragmentation.

“The industry has shifted dramatically over the past 18 months, with AI, data, and product innovation converging at speed,” James said. “We conducted a thoughtful search to find a leader who could unify AI strategy, product development, and client impact under one vision. She brings a rare blend of consulting rigor, agency experience, and deep AI expertise to lead Converged.AI forward.”

Sankar-King brings more than 25 years of experience across agencies, consultancies, and marketing services firms. Before Harte Hanks, she spent over six years as an expert partner at Bain & Company, leading global work in customer engagement, marketing transformation, and AI-driven growth. Her agency background includes senior roles at BBDO, where she served as EVP and head of marketing science, and GroupM, where she led digital media optimization and advanced analytics across North America.

In her new role, Sankar-King will oversee Havas’ analytics practice and lead Converged.AI, the agency’s AI platform that connects 23,000 people across its global network. Since its launch two years ago, the platform has served as the foundation for tools like AVA, a no-code application launched at CES in January that lets staffers build and share AI workflows across the company. James describes it as the backbone of Havas’ AI-powered organization.

Sankar-King signaled her near-term focus will be on scaling what the agency has already built. “My focus is simple: accelerate our ability to deliver intelligence-led solutions at scale,” she said. “Clients aren’t looking for more noise; they’re looking for clarity, foresight, and solutions that actually solve real business problems.”

She was pointed about where the broader industry is falling short: “Most agencies focus on surface-level AI applications — automating tasks, building shiny demos, or experimenting in silos — without addressing the systemic issues that prevent AI from delivering real value: fragmented data, disconnected workflows, insufficient infrastructure, and a misunderstanding of AI’s role in augmenting human capability.”

X will temporarily bar creators from monetizing their content through its Creator Revenue Sharing program if they share AI-generated videos of an armed conflict without clearly disclosing that it was made with AI. The move comes three days after the U.S. and Israel launched strikes against Iran, sparking a chain of violent reactions.

X’s head of product Nikita Bier made the announcement in a post on X Tuesday morning, writing: “During times of war, it is critical that people have access to authentic information on the ground.” He added that AI platforms have lowered the bar to making “content that can mislead people.” 

Today we are revising our Creator Revenue Sharing policies to maintain authenticity of content on Timeline and prevent manipulation of the program.

During times of war, it is critical that people have access to authentic information on the ground. With today’s AI technologies,…

— Nikita Bier (@nikitabier) March 3, 2026

Creators who post AI-made videos of armed conflict without adequate disclosure, Bier said, will be suspended from Creator Revenue Sharing for 90 days. Additional violations will lead to permanent suspension from the program.

Replying to one user, Bier explained that creators would need to click on the menu and select ‘Add Content Disclosures’ where they will find a ‘Made with AI’ label option.

X will be able to check whether videos are AI-generated using available metadata imbued by AI systems in combination with Community Notes, X’s crowd-sourced fact-checking tool, Bier added. 

X debuted its revenue-sharing initiative in mid-2023. Through the program, X Premium-subscribed creators or verified organizations with at least 5 million organic impressions in the past three months who have at least 500 verified followers are eligible to cash out on their content. These users to earn based on engagement—a 2024 departure from previous rules that determined payouts based on ad impressions. 

The policy update comes two days after a Wired investigation found X drowning in misinformation about the conflict in Iran. Various months- or years-old videos were presented as new, some posts included outright disinformation about specific attacks, and  AI-generated imagery and videos spread like wildfire, shared even by the Iranian newspaper Tehran Times.

X has been recently plagued with its own endemic issues related to AI-generated content, too. In late December and early January, the site was overwhelmed with nonconsensual sexualized deepfakes of real users after users prompted Grok AI, embedded within X, to produce the images. Though the platform eventually tweaked its rules, the changes were far from comprehensive, as users can still use various Grok interfaces to produce such content. 

Meanwhile, the company has also been signaling its willingness to crack down on content that might deter advertisers from spending on X. Last week, X held a webinar for advertisers—the presentation for which was obtained by ADWEEK—promoting its brand safety capabilities. X’s ad revenues are about half of what they were before billionaire Tesla boss Elon Musk acquired the platform for $44 billion in late 2022, according to data from Emarketer.

X did not comment by press time. 

Retail doesn’t stand still, and neither does Sam’s Club.

In this episode of The Speed of Culture, recorded live at CES in Las Vegas, Matt Britton sits down with Diana Marshall, EVP and chief experience officer at Sam’s Club. 

Diana shares how Sam’s Club is building a human-led, tech-powered membership model, why experience is now the primary differentiator in retail, and how closed-loop data and AI-driven personalization are reshaping acquisition, loyalty, and retail media. 

The conversation explores the shift from frictionless to effortless retail, the power of associate experience, and what the future of buying looks like in an AI-enabled world.

A 21-year Walmart veteran, Diana began her career as a replenishment trainee and has held leadership roles across merchandising, operations, and marketing before joining Sam’s Club nearly three years ago.

Her background spans supply chain, general merchandise, and enterprise innovation, giving her a rare end-to-end perspective on how large-scale retail operates. Today, she leads Sam’s Club’s experience transformation, building a human-centered, tech-powered model designed to differentiate the brand in a rapidly evolving membership economy.

Key takeaways:

[00:04:56] Experience as the Differentiator — Sam’s Club recently formalized its experience organization, a first within the Walmart enterprise. While price, assortment, and trust remain foundational, Diana explains that experience is where the brand can truly win. The vision: create effortless, personal experiences that bring joy and foster community. It’s a shift from competing on value alone to competing on how members feel across physical clubs, digital channels, and human interactions.

[00:07:04] You Can’t Build Member Experience Without Associate Experience — One of Diana’s biggest early insights: member experience cannot improve unless associate experience improves first. From onboarding to tools to leadership culture, Sam’s Club is investing in setting associates up for success. The logic is simple—empowered, supported employees create better human moments. In a tech-heavy era, the associate remains a competitive advantage.

[00:10:46] The Closed-Loop Advantage of Membership Data — Because Sam’s Club operates on a membership model, it has full closed-loop attribution across behavior, purchase, and engagement. Diana describes building unified data systems that power next-best-action engines and real-time personalization. With structured, centralized member data, AI becomes actionable allowing Sam’s Club to move from broad campaigns to highly contextual experiences at scale.

[00:18:29] Retail Media as an Experience Platform — Sam’s Club’s Member Access Platform isn’t positioned as just another retail media network. Diana describes it as a “retail experience network,” designed not just to serve ads but to improve member experience. With closed-loop measurement and personalization capabilities, MAP supports brand partners while also driving acquisition and loyalty turning media into a strategic growth engine.

[00:25:51] Leadership in the Age of AI — In a world obsessed with automation, Diana centers leadership around humanity. Her mantra—“Be kind. Be curious. Be inclusive. Tell the truth always.”—reflects how she builds teams. She emphasizes risk-taking, embracing imperfection, and leading with empathy. For her, technology scales performance, but culture scales impact.