LinkedIn is launching a new AI-powered feed ranking system that uses large language models and GPUs to analyze post content and surface more relevant updates to its 1.3 billion members.

Why we care. Understanding how LinkedIn surfaces content is critical if you want your posts — or your brand’s — to be discovered. The new system prioritizes topical relevance and engagement patterns, LinkedIn said. Posts that demonstrate expertise and align with emerging professional conversations may travel farther across the network — even without existing connections.

The details. LinkedIn rebuilt much of its feed recommendation system using large language models, transformer models, and GPU infrastructure. The overhaul centers on two systems: retrieving relevant posts and ranking them in the feed.

Unified retrieval system. LinkedIn replaced several separate discovery systems with a single LLM-powered retrieval model.

• Previously, feed candidates came from multiple sources, including network activity, trending posts, collaborative filtering, and topic-based systems.
• The new approach uses LLM-generated embeddings to understand what posts are about and how they connect to your professional interests.
• Now, LinkedIn can link related topics even when they use different terminology. For example, engagement with posts about small modular reactors could surface content about electrical grid infrastructure or renewable energy.

Ranking that follows your interests. After retrieval, LinkedIn ranks posts using a transformer-based sequential model. Instead of evaluating posts independently, the model analyzes patterns across your past interactions — including likes, comments, dwell time, and other signals.

• This helps LinkedIn detect how your professional interests evolve and recommend content that reflects those shifts.

System performance and infrastructure. The system runs on GPU infrastructure designed to process millions of posts while keeping feeds fresh.

• The architecture can update content embeddings within minutes and retrieve candidates in under 50 milliseconds, LinkedIn said.

Improving feed quality and authenticity. LinkedIn also announced updates to improve content quality:

• Cracking down on automated engagement. LinkedIn is taking action against comment automation tools, browser extensions, and engagement pods that create inauthentic conversations. These tools violate platform rules and undermine real professional discussions, LinkedIn said.
• Reducing engagement bait and generic posts. LinkedIn plans to show less content designed purely to drive comments or clicks — including posts asking people to comment “Yes” to boost reach, posts pairing unrelated videos with text to game distribution, and recycled thought-leadership with little substance.
• Helping new members personalize their feeds faster. LinkedIn is testing an “Interest Picker” during signup that lets new users choose topics such as leadership, job search skills, or career growth, helping deliver relevant content from day one.

SerpApi is asking a federal court to dismiss Reddit’s lawsuit over alleged scraping of Reddit content from Google Search, saying Reddit is trying to use copyright law to control user posts and public search results.

• The motion follows Reddit’s amended complaint filed in February.
• SerpApi says the filing still fails to show copyright ownership, circumvention of technical protections, or concrete harm.

SerpApi’s argument. SerpApi CEO Julien Khaleghy, in a blog post today, argued the lawsuit fails for several reasons:

• Reddit doesn’t own most of the content at issue. Its user agreement states that users retain ownership.
• Reddit holds only a non-exclusive license to user posts.
• The snippets cited in the complaint (e.g., dates, addresses, short fragments) aren’t copyrightable.
• SerpApi accessed Google Search pages, not Reddit itself.

DMCA. Khaleghy said Reddit claims SerpApi violated the Digital Millennium Copyright Act (DMCA) by circumventing technical protections. SerpApi disputes that claim, saying it retrieves the same search results visible to anyone who enters a query in Google. Khaleghy argued that:

• SerpApi doesn’t break encryption or bypass authentication.
• Accessing public webpages isn’t “circumvention” under the DMCA.
• Reddit is trying to enforce copyright protections it doesn’t own.
• Reddit’s privacy policy states that public posts may appear in search results.

Catch up quick. Legal fights over search scraping and AI data have intensified in recent months:

• Oct. 22: Reddit sued SerpApi, Perplexity, Oxylabs, and AWMProxy, alleging they scraped Reddit content through Google Search and reused it at scale. Reddit said it planted a “trap” post visible only to Google’s crawler that later appeared in Perplexity responses.
• Oct. 29: SerpApi said it would fight the lawsuit, calling the allegations inflammatory and defending access to public search data.
• Dec. 19: Google sued SerpApi, alleging the company bypassed bot protections and scraped licensed search features.
• Feb. 23: SerpApi asked a federal court to dismiss Google’s lawsuit, arguing Google is misusing the DMCA to restrict access to public search results.

Why we care. The case tests whether companies can extract information from Google’s search results without violating copyright or the DMCA. The outcome could affect SEO tools and AI training data.

What’s next. The court must decide whether Reddit’s amended complaint can proceed. If the judge dismisses the case with prejudice, Reddit’s claims against SerpApi in this lawsuit would end.

SerpApi’s blog post. Reddit’s Lawsuit is a Dangerous Attempt to Expand Platform Power

ChatGPT retrieves far more webpages than it cites. A new AirOps analysis found that 85% of discovered sources never appear in the final answer.

Why we care. If you want your content cited in AI-generated answers, discovery isn’t enough. Most retrieved pages never become visible to users.

Key finding. In AI answers, retrieval doesn’t equal citation. Your page can rank and be retrieved yet still lose the citation to a source that better matches the prompt or supporting context.

• This shifts optimization toward earning selection inside the AI synthesis process—not just appearing in search results, per the report.

By the numbers:

• 82,108 citations appeared in final responses.
• Only 15% of retrieved pages were cited.
• 85% of pages surfaced during research never appeared in answers.

Citation rates also varied by query type:

• 18.3% for product discovery queries
• 16.9% for how-to queries
• 11.3% for validation searches

Fan-out queries. ChatGPT often expands prompts with additional internal searches while generating an answer, creating what the report calls a “second citation surface.” Across the dataset:

• 89.6% of prompts triggered two or more follow-up searches.
• Fan-out searches expanded 15,000 prompts into 43,233 queries.
• 32.9% of cited pages appeared only in fan-out results—not the original prompt.
• 95% of fan-out queries had zero traditional search volume.

Google ranking correlation. High Google rankings strongly correlated with citations:

• 55.8% of cited pages ranked in Google’s top 20.
• Pages ranking in Position 1 were cited 3.5 times more often than pages outside the top 20.

About the data. AirOps analyzed 548,534 pages retrieved across 15,000 prompts to examine how ChatGPT expands queries and selects citations.

The study. The Influence of Retrieval, Fan-out, and Google SERPs on ChatGPT Citations

Google’s AI Overviews may be reducing traditional search clicks, but publishers still have meaningful growth opportunities in breaking news and Google Discover, according to new data from Define Media Group.

• Organic search clicks have fallen 42% since AI Overviews began expanding in Google Search, according to Define Media Group’s analysis of Google Search Console data across its portfolio of 64 sites.

Why we care. AI-generated answers are reshaping search traffic. Evergreen content is losing clicks, while real-time news coverage and Discover distribution are emerging as stronger traffic channels for publishers.

By the numbers. Across Google Search, Discover, and Google News, breaking news traffic grew 103% from November 2024 through early 2026 in the company’s dataset. Losses were concentrated in informational and evergreen content:

• Organic search traffic averaged 1.7 billion clicks per quarter from Q1 2023 through Q1 2024.
• After AI Overviews launched, traffic fell 16% immediately and never recovered.
• As Google expanded AI Overviews in May 2025, declines accelerated.
• By Q4 2025, search traffic was down 42% from the pre-AI Overviews baseline.

Discover’s role: Google Discover, which grew 30% across the portfolio, is now the main growth engine for breaking news distribution. Discover traffic rose steadily as web search traffic fell. For the first time in the dataset, Discover and web search now drive roughly equal traffic.

• The report also found a sharp increase in Discover traffic after the December 2025 Google core update, though some gains softened after Google’s February 2026 Discover core update.
• Meanwhile, Chartbeat data showed Discover was the primary driver of Google referrals to news sites as of last summer. See: Google traffic to news publishers is steady, but it isn’t traditional Search.

Why is this happening? AI Overviews appear less often for news queries than for other topics. AI Overviews appeared for about 15% of news queries — nearly three times less often than in categories such as health and science — according to Ahrefs data cited in the report.

• News queries often trigger the Top Stories carousel, which links directly to publisher articles. Searches for major developing events, such as international conflicts, typically show Top Stories rather than AI summaries.
• Define Media Group suggests Google may be avoiding AI-generated summaries for breaking news because events change rapidly, accuracy stakes are high, and generative systems can still hallucinate.

The report. BREAKING! News Thrives in the Age of AI