In a climate where most brands would give anything for a case study boasting 10x growth, PacSun recently hit the ultimate benchmark: 200,000 pairs of jeans sold via TikTok, generating $20 million in revenue. 

But as the retailer is quickly learning, creativity without compliance is just a high-interest loan.

The very numbers that should be headlining quarterly business reviews are now being served as Exhibit A in a $43 million lawsuit from Warner Music Group. Seeking $150,000 in damages for each of the 290 instances of alleged music infringement, the label remixed a bop into a multimillion-dollar invoice anthem.

PacSun is just the latest in a growing gallery of high-profile offenders, including 14 NBA teams, Johnson & Johnson, and Chili’s. For years, the industry’s unspoken mantra has been “ask for forgiveness, not permission.” But as labels pivot from “whack-a-mole” enforcement to precise legal strikes, the industry is facing a reckoning: Our creative ambitions have outpaced our operational infrastructure.

The Business Affairs bottleneck

Working at the intersection of social and production, I see the gap daily. Social teams need to post at the speed of culture, yet “Business Affairs”—the gatekeeper of rights and clearances—remains a relic of linear advertising. We’re applying a 1990s legal framework to a 2026 feed. 

Agencies are designed to spend weeks negotiating a single track for a TV spot, but a social lead needs to post a trending reel within 30 minutes. This friction has created a dangerous vacuum where “trusting the algorithm” has replaced due diligence.

The math simply doesn’t track. It can cost between $3,000 and $5,000 to clear a single commercial track with a label for one organic post on one platform. Multiply that by more than 20 influencers—all using the same trending Drake or Beyoncé snippet—and you aren’t just looking at a budget line item. You’re looking at a fiasco.

From “mute” to modern infrastructure

With nearly 85% of TikToks relying on sound and more than 44% of users buying directly through the app, silence isn’t an option. But neither is the status quo. 

A major label recently disclosed it was tracking 22,000 brands illegally using between 10 and 20 songs per month. This is a gaping structural disconnect between how we create content and how we protect the enterprise.

Fixing this requires a fundamental shift in how we source sound. “Clearinghouses”—like Cipher and other pre-licensed libraries—are moving from niceties to requirements. These platforms are the first step toward a true ecommerce model for music. However, if you’re looking for a “see it, click it, buy it” solution for a Taylor Swift track, you won’t find one.

The barriers are deeply entrenched in the industry’s DNA. A single track often has a dozen songwriters, multiple publishers, and a record label, who all must give the green light. “Most-favored-nation” (MFN) clauses—which dictate that all rightsholders must be paid the same rate—are a massive brake on automation. A “no” from one minor songwriter could empty the whole digital cart. 

Until the industry can move toward automated, pre-negotiated bundles across a brand’s social footprint, the legal receipt will continue to outpace the creative win.

Creator-led collabs

By collaborating with artists, brands can stop borrowing trends and start owning them.

Skeptics argue that bespoke collaboration can’t scale at the speed of social. But scalability lives underneath the mainstream market. Independent music creators are often sitting on vast libraries of unreleased tracks, and they are far more agile than major labels. 

By tapping into this ecosystem, brands can move beyond superficially echoing a trend and connect with niche communities. Perhaps they can even start a trend of their own.

Intentional audio increases engagement far more effectively than a generic Top 40 snippet. It’s the difference between a travel brand using Jess Glynne’s “Hold My Hand” to create a genuine phenomenon versus an ice cream brand forcing a fit with a Drake track just because it’s “hot.” 

Facing the music

Some are holding out hope that countersuits—like DSW’s argument that platforms should be held responsible for the music in their libraries—will save the day. That is a slippery slope.

The era of “accidental” infringement is over, with labels carefully watching bottom lines. So, if your agency or brand team hasn’t updated its social-audio operations in the last 12 months, you aren’t just making content—you’re making a paper trail. 

In 2026, the only thing more expensive than a global ad campaign is a “free” song.

https://www.adweek.com/social-marketing/from-trending-sound-to-43-million-legal-receipt/

https://www.adweek.com/social-marketing/pinterest-shoppable-ctv-series-turns-pins-into-purchases-on-roku/

Brands have caught on and started engaging on our platform, but that doesn’t guarantee it’ll be captured by the LLMs. The search ecosystem is still under construction, and while we don’t have all the answers yet, what we know so far is that showing up on Reddit does correlate with an increase in LLM mentions. https://www.adweek.com/media/the-secret-to-brand-visibility-in-the-llm-era/

TBPN’s newest sponsors include the New York Stock Exchange (NYSE), Shopify, Plaid, CrowdStrike, AppLovin, and others. https://www.adweek.com/media/tbpn-advertisers-shopify-plaid-applovin-nyse/

Unveiled at CES, Max Campaign aims to win performance ad dollars while offering brands more transparency. https://www.adweek.com/media/reddit-max-campaign-media-buying-ces/

The US has adopted more of a middle ground approach, essentially letting private companies decide what they wanted to do. Daymude and his co-authors wanted to investigate these markedly different approaches. So they developed a computational agent-based simulation that modeled how individuals navigate between wanting to express dissent versus fear of punishment. The model also incorporates how an authority adjusts its surveillance and its policies to minimize dissent at the lowest possible cost of enforcement.

“It’s not some kind of learning theory thing,” said Daymude. “And it’s not rooted in empirical statistics. We didn’t go out and ask 1000 people, ‘What would you do if faced with this situation? Would you dissent or self-censor?’ and then build that data into the model. Our model allows us to embed some assumptions about how we think people behave broadly, but then lets us explore parameters. What happens if you’re more or less bold? What happens if punishments are more or less severe? An authority is more or less tolerant? And we can make predictions based on our fundamental assumptions about what’s going to happen.”

Let one hundred flowers bloom

According to their model, the most extreme case is an authoritarian government that adopts a draconian punishment strategy, which effectively represses all dissent in the general population. “Everyone’s best strategic choice is just to say nothing at this point,” said Daymude. “So why doesn’t every authoritarian government on the planet just do this?” That led them to look more closely at the dynamics. “Maybe authoritarians start out somewhat moderate,” he said. “Maybe the only way they’re allowed to get to that extreme endpoint is through small changes over time.”

Daymude points to China’s Hundred Flowers Campaign in the 1950s as an illustrative case. Here, Chairman Mao Zedong initially encouraged open critiques of his government before abruptly cracking down aggressively when dissent got out of hand. The model showed that in such a case, dissenters’ self-censorship gradually increased, culminating in near-total compliance over time.

But there’s a catch. “The opposite of the Hundred Flowers is if the population is sufficiently bold, this strategy doesn’t work,” said Daymude. “The authoritarian can’t find the pathway to become fully draconian. People just stubbornly keep dissenting. So every time it tries to ramp up severity, it’s on the hook for it every time because people are still out there, they’re still dissenting. They’re saying, ‘Catch us if you dare.’”

https://arstechnica.com/science/2025/12/the-science-of-how-and-when-we-decide-to-speak-out-or-self-censor/

https://www.adweek.com/media/2026-media-predictions/

Like most tools, generative AI models can be misused. And when the misuse gets bad enough that a major dictionary notices, you know it’s become a cultural phenomenon.

On Sunday, Merriam-Webster announced that “slop” is its 2025 Word of the Year, reflecting how the term has become shorthand for the flood of low-quality AI-generated content that has spread across social media, search results, and the web at large. The dictionary defines slop as “digital content of low quality that is produced usually in quantity by means of artificial intelligence.”

“It’s such an illustrative word,” Merriam-Webster president Greg Barlow told the Associated Press. “It’s part of a transformative technology, AI, and it’s something that people have found fascinating, annoying, and a little bit ridiculous.”

To select its Word of the Year, Merriam-Webster’s editors review data on which words rose in search volume and usage, then reach consensus on which term best captures the year. Barlow told the AP that the spike in searches for “slop” reflects growing awareness among users that they are encountering fake or shoddy content online.

Dictionaries have been tracking AI’s impact on language for the past few years, with Cambridge having selected “hallucinate” as its 2023 word of the year due to the tendency of AI models to generate plausible-but-false information (long-time Ars readers will be happy to hear there’s another word term for that in the dictionary as well).

The trend extends to online culture in general, which is ripe with new coinages. This year, Oxford University Press chose “rage bait,” referring to content designed to provoke anger for engagement. Cambridge Dictionary selected “parasocial,” describing one-sided relationships between fans and celebrities or influencers.

The difference between the baby and the bathwater

As the AP points out, the word “slop” originally entered English in the 1700s to mean soft mud. By the 1800s, it had evolved to describe food waste fed to pigs, and eventually came to mean rubbish or products of little value. The new AI-related definition builds on that history of describing something unwanted and unpleasant.

https://arstechnica.com/ai/2025/12/merriam-webster-crowns-slop-word-of-the-year-as-ai-content-floods-internet/

https://www.adweek.com/media/ftc-meta-ruling-antitrust-instagram-whatsapp/

The next time you encounter an unusually polite reply on social media, you might want to check twice. It could be an AI model trying (and failing) to blend in with the crowd.

On Wednesday, researchers from the University of Zurich, University of Amsterdam, Duke University, and New York University released a study revealing that AI models remain easily distinguishable from humans in social media conversations, with overly friendly emotional tone serving as the most persistent giveaway. The research, which tested nine open-weight models across Twitter/X, Bluesky, and Reddit, found that classifiers developed by the researchers detected AI-generated replies with 70 to 80 percent accuracy.

The study introduces what the authors call a “computational Turing test” to assess how closely AI models approximate human language. Instead of relying on subjective human judgment about whether text sounds authentic, the framework uses automated classifiers and linguistic analysis to identify specific features that distinguish machine-generated from human-authored content.

“Even after calibration, LLM outputs remain clearly distinguishable from human text, particularly in affective tone and emotional expression,” the researchers wrote. The team, led by Nicolò Pagan at the University of Zurich, tested various optimization strategies, from simple prompting to fine-tuning, but found that deeper emotional cues persist as reliable tells that a particular text interaction online was authored by an AI chatbot rather than a human.

The toxicity tell

In the study, researchers tested nine large language models: Llama 3.1 8B, Llama 3.1 8B Instruct, Llama 3.1 70B, Mistral 7B v0.1, Mistral 7B Instruct v0.2, Qwen 2.5 7B Instruct, Gemma 3 4B Instruct, DeepSeek-R1-Distill-Llama-8B, and Apertus-8B-2509.

When prompted to generate replies to real social media posts from actual users, the AI models struggled to match the level of casual negativity and spontaneous emotional expression common in human social media posts, with toxicity scores consistently lower than authentic human replies across all three platforms.

To counter this deficiency, the researchers attempted optimization strategies (including providing writing examples and context retrieval) that reduced structural differences like sentence length or word count, but variations in emotional tone persisted. “Our comprehensive calibration tests challenge the assumption that more sophisticated optimization necessarily yields more human-like output,” the researchers concluded.

https://arstechnica.com/information-technology/2025/11/being-too-nice-online-is-a-dead-giveaway-for-ai-bots-study-suggests/