AI Made Content Cheap. Brands Are Turning to Newsrooms to Stay Heard.

Several of September's biggest marketing stories have something in common. LinkedIn reach is falling, AI engines are citing third parties instead of brands, and companies are writing formal rules about how employees can use ChatGPT. Each one is a symptom of the same shift. Content has become cheap to produce, and most of it now struggles to earn attention.

The supply of content has doubled

Researchers can now put numbers on the flood. Graphite sampled 55,400 English-language URLs from Common Crawl and found that primarily AI-generated articles have made up about half of all new articles since early 2025. According to Graphite's full report, AI-written articles briefly passed human ones at 50.9% in the fourth quarter of 2025 before dipping to 49.9% in the first quarter of 2026. Pew Research Center used a different method in its August analysis and found signs of AI authorship in more than a third of pages published since ChatGPT launched.

The share has stopped growing, and Graphite has a theory about why. Its researchers believe publishers found that mostly AI-written articles don't perform well in search. Volume went up, but the extra volume mostly made ordinary content harder to find.

Companies are running into the same problem internally. In research from BetterUp Labs and Stanford, 40% of desk workers said they had received low-quality AI output from a colleague in the past month, and 42% said they trusted that colleague less afterward. Harvard Business Review gave the problem its name, "workslop." One response has been a wave of published AI writing policies. Clay co-founder Varun Anand made his company's policy official across the business in August, and his LinkedIn post about it drew nearly 10,000 reactions. For many buyers, a public promise that people wrote your content has become a selling point.

Distribution is getting more expensive

While the supply of content grew, the channels brands use to reach people got narrower. AuthoredUp data summarized by Botdog shows LinkedIn reach down 47% year over year and video reach down 72%. TryOrdinal data cited by Linkboost puts the decline in company page organic reach at 60% to 66% between 2024 and 2026.

What reach remains goes to a small group of accounts. In Executive Presence's 2026 CEO study, seven executives, 13% of the group, collected 47% of all impressions, as Bill Hartzer noted in his review. Over the same period, LinkedIn revenue grew 12% in Microsoft's June quarter, driven mainly by its advertising business. State of Brand reported this week on the paid tools LinkedIn has rolled out to help brands buy back the reach they lost.

AI search adds a second squeeze. AirOps' 2026 State of AI Search report found that 85% of brand mentions in AI answers come from third-party pages rather than a brand's own site. G2's research found that 51% of B2B software buyers now start research in an AI chatbot more often than on Google. When a buyer asks an AI engine about your company, the answer is mostly built from what other people have published about you.

Put together, brands are dealing with three problems at once. Their content looks like everyone else's, their social reach costs money to recover, and AI tools give little weight to what brands say about themselves.

Becoming the source

Melissa Rosenthal, co-founder of Outlever, which publishes State of Brand, has made this argument in podcast appearances throughout the year. Most recently she joined Dave Gerhardt on the Exit Five podcast. In her post about the episode, she summed up her view this way: "You don't win by 'doing content.' You win by building a media company."

Rosenthal was an early employee at BuzzFeed, a company that grew on other platforms' distribution and suffered when those platforms changed course. In a Superpath AMA this summer, she said the main lesson she took from that period is that "the rails are never yours." Her approach for B2B companies comes down to three steps.

1. Build a news source instead of a blog. Rosenthal argues that the typical B2B blog was written for a search algorithm and never built a real readership, a trend State of Brand has covered in its reporting on the collapse of the B2B blog. She recommends starting small. Set aside a section of your existing site where your team can publish news and commentary quickly, and focus on topics your industry cares about that also connect to your business.

She also believes a small market can be an advantage. A company trying to reach 5,000 people has an easier path to becoming their main news source than one chasing 500,000, because niche audiences are often underserved. Outlever built Benefits Brief for health benefits company SureCo to cover the ICHRA market for insurance brokers, and she says it became widely discussed in that industry within a few months.

This kind of reporting also matters for AI search. A Search Engine Land study, cited in State of Brand's analysis of AI citations, found that 52.2% of the passages AI engines cited contained original or proprietary data. Engines are far more likely to cite original reporting than repackaged content.

2. Build the publication around people. An AI model can write an opinion about your market, but it cannot interview your customers. On The Changemakers podcast, Rosenthal described Outlever's approach as featuring subject matter experts, customers and partners rather than executives alone. Her team also selects interview subjects from a client's target accounts. She says that produces demand generation and account-based marketing results within weeks.

Interview-based stories also avoid a problem most content teams know well. Customers who would never approve a formal case study are often happy to talk about what is happening in their industry. The distribution data points in the same direction. LinkedIn now favors posts from individuals, and AI engines lean on named experts and employees more than on brand copy. A publication built on interviews supplies both.

3. Publish at the speed of news and measure like a publisher. Rosenthal considers timeliness the main difference between a newsroom and a content calendar. She calls newsjacking, meaning quick reactions to industry news, the most effective approach on LinkedIn right now. She also warns against building everything around one platform's current preferences, because those preferences change. In her model, LinkedIn is a way to distribute the publication, and the publication itself lives on a site the company owns.

On measurement, she tells marketing leaders to watch early signals before revenue shows up. Are the right people reading? Are they returning and sharing? Are the right customer voices being featured? She has seen many companies shut these programs down while waiting for precise attribution that takes time to arrive. The financial argument rests on comparing owned media with paid media. Ad spending stops producing results as soon as the budget runs out. A publication keeps attracting readers over time, and the cost per reader tends to fall as it grows. State of Brand has examined how companies are staffing these teams and why some brand newsrooms fail.

The results so far

State of Brand is Outlever's own test of the approach. Rosenthal says the publication reached millions of visitors within three months. Most of that traffic now comes from sources Outlever doesn't pay for, including syndication, AI search and email newsletters.

Few large companies are competing for this ground yet. A McKinsey survey found that only 16% of Fortune 500 consumer-brand CMOs systematically track how their brands perform in AI search. Most teams still judge content by how much they publish.

For marketing leaders planning 2027 budgets, a more useful test is whether people in your industry would notice if your content stopped appearing.

Disclosure: State of Brand is published by Outlever, which is mentioned in this article. Melissa Rosenthal is Outlever's co-founder.