Welcome to today's edition of AI Marketing Geek, your daily dose of the latest AI developments, tools, trends, and actionable insights shaping marketing in 2026. As AI shifts from experimental hype to pragmatic, ROI-driven reality, we're seeing agentic AI take center stage. Autonomous agents are handling media buying, personalization, and commerce, which is wild to watch.

Something clarified in February that did not get enough attention at the time. The AI industry has divided into two explicit camps on how to make money, and the line is drawn around advertising.

Perplexity pulled the plug on its advertising experiment just over a year after launching it. The company was one of the first AI search platforms to test sponsored answers, placing them beneath chatbot responses in 2024. Company executives now say ads risk making users "suspicious of everything."

The revenue numbers make this decision more interesting. Perplexity generated approximately $20,000 in advertising revenue in 2024 out of $34 million in total revenue, representing less than 0.1% of total revenue. That limited financial dependence on advertising made the decision to abandon it strategically easier, as the company wasn't giving up a substantial existing revenue stream. Instead, the company has reached $200 million in annual recurring revenue on subscriptions alone, nearly five times year-over-year growth.

Anthropic has made its skepticism equally explicit, running a Super Bowl campaign with the tagline "Ads are coming to AI. But not to Claude," positioning its chatbot as a distraction-free alternative. Google already serves ads within its AI Overviews and AI Mode features on search, but has drawn a line at placing ads directly within Gemini itself.

OpenAI is going the opposite direction. ChatGPT launched ads for free and low-cost tier users in February at a $60 CPM with a $200,000 minimum commitment, and has since brought Criteo into the ecosystem to scale programmatic demand.

The strategic bet each side is making is actually quite different. OpenAI is betting it can maintain user trust while serving ads, the same bet Google made with search 25 years ago and won. Perplexity and Anthropic are betting that AI answers are fundamentally different from search results, that users won't tolerate the ambiguity of wondering whether a recommendation is paid. One Perplexity executive framed it simply: "We're in the accuracy business, and the business is about delivering the truth, the right answers."

For marketers, the split matters in two ways. First, it shapes where you can actually buy AI-native ad inventory, and at what cost. Right now that means ChatGPT and Google's AI surfaces, not Perplexity or Claude. Second, it shapes how your brand is perceived inside each AI ecosystem. Platforms that prioritize trust over revenue may actually provide better organic visibility for brands that earn it, because the answer will not be competing with a paid placement sitting directly above it.

In early February, ByteDance released Seedance 2.0, an AI video generation model that immediately drew cease-and-desist letters from Disney and Paramount. That legal drama is the headline, but the underlying capability story is what matters for marketers.

Seedance 2.0 utilizes the latest generative AI video capacity to produce more refined results than many other models, with more lifelike, realistic movements and less of the moving-underwater quality that plagued earlier tools. It is currently available to some users of ByteDance's Jimeng AI platform, and it seems likely it will eventually make its way to TikTok and CapCut, which already have a range of generative AI tools available to assist creators and marketers.

The technical leap is real. Seedance 1.0 topped out at roughly 5 to 8 seconds of coherent video. Seedance 2.0 pushes that to approximately 20 seconds while maintaining temporal consistency throughout the clip. The model also incorporates physics-aware training that makes motion substantially more believable: gravity works, fabrics drape correctly, fluids behave like fluids, and object interactions look believable rather than uncanny.

One creative advertising agency owner put it plainly. Billy Boman, who runs a creative advertising agency in Stockholm that produces AI-generated content and has used Seedance 2.0, said: "Back in 2023, it was difficult to get someone to run or to walk. Any type of realism was limited to very short clips. Now I can do anything. It has been nothing short of exceptional, the technological advancements."

The Hollywood copyright problems are worth noting for practical reasons, not just legal ones. Shortly after launch, realistic clips based on real actors, TV shows, and films went viral across the internet. Disney sent ByteDance a cease-and-desist alleging infringement of its characters including Spider-Man, Darth Vader, and Grogu, while Paramount followed with its own letter covering Star Trek and South Park. As of March 2026, access to Seedance 2.0 is restricted to existing users of ByteDance's domestic Chinese apps.

For marketers, the availability question is the short-term limiter here. International access through CapCut and other platforms is coming but not yet fully open. But the speed of progress is worth tracking closely. The gap between "AI video looks fake" and "AI video looks like something a brand would actually run" closed faster than almost anyone predicted. The tools that will be broadly available to marketing teams by the end of 2026 will be substantially more capable than what existed six months ago.

Tools and Tips: How to Use AI Video Tools Without the Copyright Risk

The Seedance story above is a useful reminder that the legal guardrails around AI video are still being worked out in court. Here is how to use these tools without putting your clients or your own firm at risk.

The simple rule: only use AI video tools with content you own or have licensed. That means original photography, stock footage you have licensed rights to, product shots, brand assets, and footage you have shot. Do not feed in clips from films, TV shows, competitor ads, or music videos as reference material, even for stylistic inspiration. The model may replicate elements in ways that cross the line.

For law firm clients specifically, this matters more than you might think. Generating video that inadvertently uses a recognizable face or copyrighted piece of music creates real liability exposure. Before using any AI video tool in client work, read the platform's terms around commercial use and IP ownership. Most tools now have clear commercial use licenses, but they vary in what they permit and what they disclaim.

The safest workflow right now is to use AI video for motion graphics, product demos, text animation, and abstract visual content rather than anything featuring realistic human likenesses or recognizable characters. That keeps you well inside the guardrails while still getting the speed and cost advantages the tools offer. As the legal landscape settles, those boundaries will likely expand. For now, stay conservative and document the inputs you used if anything ever comes into question.

2026 is the year AI moves from "cool demos" to real and measurable business use cases. Expect more agentic tools, voice-powered targeting, and a focus on privacy-first data.

Stay ahead of the game by treating your AI reputation like your website. It would behoove you to make your brand easy for agents to understand and cite.

What AI experiment are you running this week? Reply and let me know. I'll feature top stories in future editions! (Everybody loves a good story)

Stay sharp,
Jon
@mistersterling
Chief AI Marketing Geek