AI-generated posts now dominate LinkedIn. Reach is the casualty.

Pangram's estimate that 41% of LinkedIn posts now show signs of AI generation is not a commentary on writing tools. It is a measurement of how thoroughly one platform's content supply has been debased.

Social Media Today reports the finding, citing Pangram's detection methodology across the platform's public post corpus. The number matters less as a precise count than as an order of magnitude: four posts in ten are written by something that has never had a client, lost a deal, or sat through a quarterly review. They read accordingly.

The mechanism is straightforward. LinkedIn's feed algorithm scores content partly on engagement signals in the first hour after posting: comments, saves, and shares from accounts with relevant professional graphs. AI-generated posts tend to produce a specific response pattern. They attract low-cost reactions from people who scroll fast and click the thumbs-up, but they rarely generate the substantive comments, direct messages, or profile visits that indicate a reader found the post genuinely useful. The algorithm is not fooled forever. Posts that accumulate likes but no comments are progressively down-ranked; posts that generate replies from senior professionals in adjacent roles get amplified. AI content, optimised for plausible prose rather than earned argument, earns the wrong kind of attention.

The authority gap is widening in real time

For B2B brands and their executives, the 41% figure changes the competitive arithmetic. When most content on a feed is interchangeable, distinctiveness becomes the scarce good. A chief risk officer at a major industrial group who posts a two-paragraph observation grounded in a specific project failure will now stand out against a feed clogged with AI summaries of industry reports. The contrast is doing work that good writing alone used to do.

This is particularly consequential for institutions where credibility is load-bearing: multilateral organisations, standards bodies, financial services firms, and policy institutions. A senior official at a UN agency or an IEEE standards committee chair is not competing on volume. They are competing on signal. One post per week that reflects actual institutional knowledge, specific enough to be unreproducible by a language model, is worth more reach than five polished AI drafts. It is also, in the current environment, more visible, because it is rarer.

The implication for company pages is more uncomfortable. Many B2B brands have accelerated posting frequency using AI assistance, chasing the conventional wisdom that more posts equals more reach. On a feed where 41% of content already reads the same, that logic inverts. Frequency without distinctiveness now suppresses authority rather than building it. LinkedIn's own internal data, shared at various partner briefings, has consistently shown that content triggering "save" behaviour outperforms content triggering "like" behaviour in sustained distribution. AI posts do not get saved. They get scrolled.

There is a subtler problem too. LinkedIn's relevance scoring uses profile visits and connection requests as downstream indicators of post quality. Executives who post content that reflects genuine expertise, including contrarian positions, specific data, or professional experience that could not have been generated from a training set, attract the kind of profile visits that convert into qualified conversations. That chain from post to pipeline depends entirely on the post being identifiably human in origin and perspective.