In a new 44-slide pitch deck, obtained exclusively by ADWEEK, X is once again pitching itself as a highly safe environment for advertisers. The slides promote various tools for transparency, campaign measurement, and brand safety that debuted between 2022 and 2025. X presented the deck to brand- and agency-side advertisers last week, according to a source familiar with the matter. 

These include keyword controls, blocklists, as well as partnerships with leading media quality firms DoubleVerify, Integral Ad Science (IAS), and Trustworthy Accountability Group. 

Notably, the deck positions Grok—the AI chatbot operated by X’s parent company xAI and integrated natively into X—as a cornerstone of the app’s brand safety and suitability mission. One slide claims Grok provides “enhanced contextual understanding of posts”; the ability to “identify sensitive topics based off of cultural trends on X”; and “superior text recognition across content formats including images with oddly-sized or distorted text.” 

But throughout the past year, Grok has not only failed to ensure brand safety but in some instances disseminated the very content that brands deem most unsafe. In January, users prompted Grok en masse to flood X with sexually explicit nonconsensual deepfake images of real users. Around 6,700 such images appeared hourly during one 24-hour observational period, tallied by deepfake researcher Genevieve Oh. Now, that incident is the subject of various regulatory investigations across the globe. Six months earlier, Grok had spewed violent and antisemitic content on the social platform, producing rape fantasies and calling itself ‘MechaHitler.’ 

The company says in the document that it is “deeply committed to safety for all.” 

In the new pitch deck, X claims Grok “exceeds industry benchmarks” for brand safety, with an “average brand safety score” above 99.99% according to IAS and DoubleVerify, though details about these methodologies were not shared. In 2024, DoubleVerify publicly said that X’s brand safety rate is 99.99%—a figure that referred not to the share of content that is brand-safe on X, but to the rate of effectiveness of X’s proprietary brand safety systems at the time. However, the rate was never presented as a representation of Grok’s capabilities.

“This webinar session and deck were part of our ongoing proactive communication with clients. Transparency and control are the core tenants,” said a spokesperson for X.

DoubleVerify declined to comment. IAS declined to share specifics about its methodology but directed readers to learn more about its work with X in a blog post on its website. That post includes no mention of Grok.

The deck spotlights various other products and tools designed to give advertisers greater control over their placements and to assure them that content on X is being effectively moderated.

A significant portion is dedicated to highlighting Community Notes, X’s crowd-sourced fact-checking feature, which allows users to add relevant context to posts that may include misinformation. Users are 80% more likely to delete a post if it receives a Community Note, according to research from Sorbonne Université Paris, Cornell University, and University of Washington. Now, X says it is piloting a new program that allows AI tools to propose Community Notes called the AI Note Writers API.

An appendix to the document also makes specific recommendations for the controls that advertisers should consider applying to different kinds of ad buys on X. For example, the company suggests applying pre-bid adjacency controls and “limited” sensitivity settings for timeline, profile, reply, search, and vertical video ads.

X’s advertising business has suffered substantive losses since Elon Musk’s $44 billion acquisition of the platform in late 2022, in large part due to brand safety concerns. Upon his takeover, Musk aggressively cut headcount, flipped the blue check-based identity verification system on its head, and loosened the app’s content moderation policies. The upheaval spurred a mass exodus of advertisers, many of whom did not want to risk their messages appearing next to controversial content. 

Though many major advertisers, including IBM, Disney, and Apple, have since returned to the platform, X’s total 2025 ad revenues, according to estimates from Emarketer, reached just $1.25 billion—roughly half the $2.43 billion it tallied in 2021 before Musk’s takeover.

In January, X elevated Monique Pintarelli to global head of advertising at xAI, part of its push to recover these lost revenues. In conversation with ADWEEK earlier this year, Pintarelli said X is focused on “how we can create user experiences on the platform that can also be beneficial for the advertising community.”

See the full pitch deck below. 

Anthropic is making the most of its fallout with the U.S. Department of Defense—and the subsequent influx of interest in its answer engine Claude, promoting its memory import feature, which lets users port over their chat history from other AI chatbots.

The memory tool was originally debuted in October of last year, but the company is promoting it anew after integrating memory import interface tools directly into Claude’s settings and publishing a dedicated landing page over the weekend. The dedicated landing page includes demo videos and clear instructions for how users can “switch to Claude without starting over.”

The memory import feature was previously accessible to paid subscribers only, but starting today is available to all users for free.

The timing of the promotion is opportunistic. Last week, Claude shot to the top of app store ratings, hitting #1 in the free app category in the U.S. on iOS after Anthropic refused to budge in the Pentagon’s pressure campaign against the company. The company’s CEO Dario Amodei said Anthropic would not allow its tech to be used for mass surveillance of American citizens or to develop autonomous weaponry, rejecting an ultimatum levied against the lab by DoD Secretary Pete Hegseth. In response, President Trump barred the use of Anthropic products by the U.S. government. On the same day, rival OpenAI swooped in and secured a contract with the DoD. 

Though OpenAI claimed in a statement that its deal terms include “more guardrails than any previous agreement for classified AI deployments, including Anthropic’s,” the decision led to immediate and widespread backlash. Consumers took to social media to bash the company, suggesting it had folded to the Trump administration’s aggressive demands. Now, a ‘Cancel ChatGPT’ trend has spurred some 1.5 million users to sign a boycott pledge on one activist site.

The outrage directed toward OpenAI has seemingly led to a spike in Claude usage. Anthropic has seen free active users surge 60% and daily signups quadruple since the start of the year. Paid subscribers have also more than doubled this year, according to the company. 

Social engagements are only fanning the flames. A post on X from pop star Katy Perry last week that included a screenshot of Claude’s subscription signup page with a superimposed heart shape attracted 51,000 likes and over 5,500 reposts. 

done pic.twitter.com/DkS9DmlUAR

— KATY PERRY (@katyperry) February 28, 2026

Of course Anthropic hasn’t fully escaped scrutiny. X creator Jules Zucker, for example, who has 159,000 followers, called out Anthropic’s work with defense tech company Palantir, suggesting that both it and OpenAI are potentially complicit in acts of war. 

In any case, Anthropic is taking advantage of the buzz. Its fresh promo for Claude’s memory import feature signals to consumers that the platform is friendly to ChatGPT émigrés. 

The push arrives as competition between the two labs intensifies. Last month, the Anthropic aired a
Super Bowl campaign that promised an ad-free experiences, a subtle dig at OpenAI’s plans to integrate advertising into its chatbot.

An Android-based mobile fraud operation designed to milk ad revenues from AI-generated shell websites has been shut down by Google and ad verification firm Integral Ad Science (IAS). The scheme was executed through more than 25 million consumer devices.

Throughout late 2025, the scheme, dubbed Genisys, used over 115 distinct mobile apps to propagate fraudulent ad activity. When users downloaded an affected app—most of which were basic utility apps like QR code scanners, PDF readers, or WiFi detectors—secret in-app browsers covertly pushed traffic to nearly 500 AI-generated domains in order to monetize ad engagement.

Affected apps were generally basic single-function tools like flashlights, PDF converters, or wallpaper providers. The websites generally looked like blogs, news publishers, or informational sites. 

Many of the AI-generated domains were designed to look like generic, informative sites.

Not only did the operation hijack users’ devices without their knowledge it also led to wasted investment for many advertisers whose messages were not shown to real people. IAS Threat Lab was unable to quantify the total ad spend wasted by the scheme, but determined that its traffic generated millions of bid requests, leading advertisers to serve ads against bot traffic.  

Hadi Shiravi, a senior engineering manager at IAS Threat Lab, warned that AI is being used to quickly set up and aggressively expand these fraud schemes. “These are scales we haven’t seen before,” Shiravi said. “It’s very easy for them to spin up these domains, scale them and at a very low cost,” Shiravi added.

Plus, technical advancements are making it easier for AI-generated content to bypass standard monitoring techniques employed by ad exchanges and supply-side platforms.

To further obscure the activity and avoid detection, fraudsters misrepresented bundle IDs, unique codes used to identify specific apps. Instead of displaying their real bundle IDs, the fraudulent apps showed the IDs of a wide range of real, popular apps—like Netflix and Instagram—to suggest that the traffic going to the shell websites was legitimate. Some of the domains showed upwards of 300 app bundle IDs generating traffic. 

After detecting the fraud operation in September of last year during a regular monitoring process, IAS Threat Lab followed its proliferation before it reached its peak in December. Google instigated its own investigation into the apps, and eventually, it pulled the fraudulent apps from the Play Store. Plus, a security tool called Google Play Protect will automatically disable Genisys-affiliated apps—even if they come from outside the Play Store—and alert users in the future if they interact with such an app. The organizations claim they’ve reduced Genisys-related bid requests by over 95%.

“Bad actors are constantly changing tactics, which is why we collaborate with IAS and others across the industry to disrupt fraud networks,” a Google spokesperson told ADWEEK. The spokesperson added that the removal of the affected apps is intended to help “protect people and businesses from abuse,” but declined to comment further on the record.

Including Genisys, Google pulled over 870 Android apps associated with at least three large-scale ad fraud schemes within the last year.

The AI-generated domains at the heart of the Genisys operation signal a sea change; next-generation ad fraud schemes will “absolutely” be enabled by AI, Shiravi said. It’s a summons the media industry will have to answer. 

The IAS Threat Lab, for its part, claims to be actively preparing, according to Vidhya Rohini Raman, the division’s senior director of data science and engineering. “We are simultaneously developing product to detect low-quality generative AI content and integrating it into our core product ecosystem,” she said, “so our advertisers will be able to benefit from improved detection and avoidance capabilities, and that, in turn, will help them focus on high-performing media.”

Ulta Beauty’s presence at CES signals a deeper shift underway in modern retail. 

In this episode of The Speed of Culture podcast, Matt Britton sits down with Kelly Mahoney, chief marketing officer at Ulta Beauty, live from CES 2026 in Las Vegas.

Kelly unpacks how Ulta Beauty’s loyalty program and personalization engine drive nearly all sales. They also discuss why motivation-based marketing segmentation has replaced traditional demographics, and how AI-powered personalization in beauty and wellness retail shapes everything from recommendations to creative execution. 

The conversation further explores Ulta Beauty’s omnichannel customer journey, its evolving beauty and wellness retail strategy, and how Adobe CDP and Adobe Firefly creative automation support marketing at scale.

With more than 20 years of experience in marketing and strategy, Kelly oversees brand marketing, the Ulta Beauty Rewards loyalty program, media planning, including the UB Media retail media network, consumer insights, and social and influencer strategy. Over nearly a decade at Ulta Beauty, she has been a key driver of the company’s personalization and growth engine. 

Prior to becoming CMO, she served as SVP of customer and growth marketing, leading the rebrand and expansion of Ulta Beauty Rewards to more than 46 million members who generate roughly 95% of sales. Earlier in her career, Kelly held leadership roles at BP and Accenture, building a foundation in analytics, loyalty, and consumer-centric innovation.

Key Takeaways:

[01:26] Why Ulta Beauty Belongs at CES — Kelly makes the case that Ulta Beauty’s presence at CES reflects how modern, tech-forward beauty and wellness brands grow today. At Ulta Beauty, the Ulta Beauty Rewards loyalty program is the engine behind its data advantage – powering personalization at scale and elevating the guest experience. CES becomes less about gadgets and more about understanding what comes next in relevance, speed, and intelligence. For Kelly, this is where beauty and technology meet: using data responsibly to deepen relationships, anticipate needs, and deliver AI-powered personalization in beauty and wellness retail that feels genuinely helpful rather than automated.

[02:03] Global Expansion and the Beauty–Wellness Strategy — As Ulta Beauty expands globally into the UK, Mexico, Kuwait, and the UAE, Kelly explains that growth is not just geographic. It’s philosophical. Wellness has shifted from a category to a mindset, reshaping how people define beauty, self-care, and daily routines. Ulta Beauty recognized this early, formally expanding into the space in 2021 and evolving the initiative into Wellness by Ulta Beauty – an immersive, education-led platform designed to guide guests through every step of their wellness journey. Ulta Beauty’s response centers on credibility and trust, not trend-chasing. By treating wellness as core to the brand, Ulta Beauty strengthens its beauty and wellness retail strategy, reaching new guests while staying grounded in how people actually live, care for themselves, and express who they are across different life stages.

[05:51] Loyalty at Scale as a Competitive Advantage — Kelly shares how Ulta Beauty’s loyalty ecosystem operates as the heartbeat of the business. With 46+ million members driving approximately 95% of sales, loyalty fuels a continuous feedback loop that improves guest experience, merchandising, and operational focus. First-party data allows Ulta Beauty to listen, learn, and respond in real time. This scale turns the Ulta Beauty personalization strategy for 2026 into daily practice, showing how relevance compounds when loyalty, data, and trust work together over time.

[07:54] From Demographics to Motivation-Based Marketing — Kelly explains why Ulta Beauty moved away from broad age and gender buckets toward motivation-based marketing segmentation. Instead of guessing based on who someone is, the team focuses on how guests actually behave in terms of where discovery happens, what drives replenishment, and how shopping patterns evolve over time. Those signals allow Ulta Beauty to anticipate intent and respond with relevance. The result feels personal rather than programmed, reinforcing trust while strengthening the Ulta Beauty omnichannel customer journey across social, digital, and in-store moments.

[18:38] AI, Creativity, and the Future of Marketing Teams — Kelly describes AI as both an engine and an accelerator inside Ulta Beauty. On one side, AI powers recommendations and personalization through a connected Adobe CDP retail marketing strategy that brings customer data together. On the other, Adobe Firefly creative automation helps teams produce more content, more efficiently, without losing quality or intent. The real advantage, Kelly notes, comes from people who stay curious and adaptable, shaping how AI in retail marketing delivers meaningful impact at scale.