What happened in SEO & AI search news this month?
Each month, Lumar’s in-house tech SEO experts hand-pick some of the SEO industry’s top news stories from across the web to keep you up-to-date on all things SEO and website optimization.
For our February 2026 SEO & GEO/AEO news roundup, the top headlines include:
- Google’s February 2026 Discover Core Update rolled out this month with an aim to prioritize more locally relevant content and sources in users’ feeds.
- A new Google WebMCP protocol launches to help AI agents interact with websites.
- Google says it will make links in AI Overviews and AI Mode more prominent.
- AI performance insights are now available in Bing Webmaster Tools.
- Cloudflare launches “Markdown for AI Agents” feature.
- The AI chatbot ad debate heats up as OpenAI begins testing ads on ChatGPT while Claude and Perplexity pull away from ad models.
- (and more!)
Dive into the details behind the SEO and GEO/AEO headlines this month below…
But first, did you know you can now optimize for AI search and LLM brand mentions with Lumar?
Lumar’s GEO tools for AI search optimization provide data-driven insights to help you optimize your website for AI-driven search visibility, earn more AI citations, and improve brand visibility across LLMs.
Google rolls out its February 2026 Discover Core Update
Google’s Discover Core Update launched on February 5th, 2026, with an expected rollout time of around two weeks. The update focuses solely on Google Discover, so broader rankings and visibility should remain unchanged.
The update aims to make Google Discover more useful by prioritizing locally relevant content from websites based in a user’s own country. Google also promises a reduction in clickbait being surfaced and a shift towards more in-depth, timely content from publishers with demonstrated expertise on specific topics.
(Source: Google Search Status Dashboard)
Google announces WebMCP protocol to help AI agents interact with websites
After last month’s UCP (Universal Commercal Protocol) announcement for AI e-commerce, Google has now also announced its WebMCP (Web Model Context Protocol) to help AI agents better interact with websites. Proposed as a new web standard, WebMCP helps site owners expose structured “tools” to LLMs, enabling AI agents to perform previously complex website actions more quickly and effectively.
Per the Chrome for Developers blog announcement:
“As the agentic web evolves, we want to help websites play an active role in how AI agents interact with them. WebMCP aims to provide a standard way for exposing structured tools, ensuring AI agents can perform actions on your site with increased speed, reliability, and precision.”
These structured tools can replace the process of AI trying to read pages, click buttons, or scrape content. Instead, bots will be able to use these tools to simplify everything from searching for products to filling out forms. The protocol is currently available for early preview, but we’ll continue to monitor its development closely.
(Source: Google: Chrome for Developers Blog)
AI performance insights come to Bing Webmaster Tools
Microsoft has officially rolled out AI Performance in Bing Webmaster Tools. It’s the first time site owners have direct insight into where and when their content is being cited within AI search, with data coming directly from Microsoft Copilot and Bing’s AI summaries.
The new Bing dashboard surfaces five core AI performance metrics, including the number of citations per page and AI visibility trends over time, much like the organic traffic data that currently appears in GSC. It signals a massive shift towards GEO/AEO tracking from a major industry player, empowering sites to validate which pages contribute to AI answers, identify content gaps, and refine their pages for stronger AI visibility.
(Source: Microsoft)
Google updates its documentation on HTML file size limits
Google has updated its documentation on HTML file size limits, clarifying that Googlebot for Search uses a 2MB limit for HTML. Previously, 15MB was listed as the default ceiling for all of Google’s crawlers.
The update caused a stir in the SEO world, but it doesn’t actually seem linked to a change in crawler behavior. In its document change log, Google describes this update as one that simply allows it to be more precise about Googlebot’s limits. It’s also worth noting that webpages reaching this 2MB HTML threshold are incredibly rare, with the vast majority of pages never even coming close to this limit.
(Source: Search Engine Journal)
Cloudflare launches “Markdown for AI Agents” feature
Cloudflare has launched Markdown for Agents, a new feature that lets AI systems request content in Markdown instead of traditional HTML. The switch will happen automatically whenever an AI bot includes an ‘Accept: text/markdown’ header in its request.
Markdown can be notably much lighter than HTML and is therefore considered by Cloudflare as easier for AI bots to read. In Cloudflare’s own example, a page that took over 16,000 tokens in HTML drops to just 3,150 tokens in Markdown; roughly an 80 % reduction in token usage. The result is lower AI processing costs and a new way to surface content so it’s clearer for bots and LLMs to understand. However, Markdown has recently been criticized by representatives from both Google and Bing, who argue that it will unnecessarily increase crawl load and add noise to the already ‘messy’ web infrastructure.
(Source: Cloudflare)
Google says links in AI Overviews and AI Mode will become more noticeable
Google has pledged to make link indicators within its AI Overviews and AI Mode more prominent and descriptive. The change is happening across both mobile and desktop search interfaces and aims to help address concerns around publisher visibility in AI responses.
Desktop users will see an additional change in how links are displayed, with link icons showing a grouped list of links when hovered over. These links come complete with site names and icons, alongside article previews. Announcing the change on X, Google’s VP of Product says the new UI proved to be more engaging in user tests.
(Source: Search Engine Journal)
ChatGPT is now processing 2.5 billion prompts per day, but its click-through rate is roughly 96% lower than Google
New research from Ahrefs shows that ChatGPT now processes roughly 2.5 billion prompts per day, compared to Google’s estimated 13.7 billion daily search queries. This has quietly made ChatGPT a bigger destination than Bing for search‑like activity.
Despite the large number of prompts currently being processed daily by ChatGPT, Google still sends about 190x more traffic to sites than ChatGPT, with ChatGPT’s estimated click‑through rate roughly 96% lower than Google’s. It highlights a fundamental difference in how the two search functions work. Google is still largely about connecting users to websites via blue links, while ChatGPT often keeps users inside the conversation.
(Source: Ahrefs: “ChatGPT Has 12% of Google’s Search Volume but Google Sends 190x More Traffic to Websites”)
AI configuration features officially launch in Google Search Console
Google’s AI-powered configuration feature has now rolled out in GSC for all users, making it easier to build customised Performance reports using natural language prompts. In the Performance section of the tool, site owners can now describe the filters and settings they’d like to apply. Search Console will then automatically set up the dashboard based on the specified dates, queries, and metrics. The GSC AI configuration feature was first announced by Google in December of last year.
(Source: Search Engine Land)
OpenAI begins testing ads in ChatGPT
OpenAI has now started testing ads within the ChatGPT interface. At present, the change is live only for logged-in adults in the US who are using a free or low-cost subscription tier. The ads aim to fund broader access to the tool while keeping the chatbot experience the same. OpenAI states that the ads do not influence the responses provided by ChatGPT, and conversations are not shared with advertisers.
(Source: OpenAI)
Perplexity & Claude will not serve ads on their AI platforms, in contrast to ChatGPT
In contrast to OpenAI’s current ad testing on ChatGPT, Perplexity is pulling back from advertising on its AI platform, and Anthropic has committed to not serving ads on Claude, signaling a broader industry debate over how generative AI platforms should monetize without undermining the perceived “truthfulness” of responses.
According to reporting in The Verge, Perplexity’s leadership concluded that incorporating ads could undermine user trust in answers, noting concerns that users might “start doubting everything” if commercial ad content was intermingled with AI responses. The company has phased out ads that it had begun testing on Perplexity back in 2024 and is instead prioritizing paid offerings aimed at professional and enterprise users, such as finance, legal, and healthcare sectors.
This puts Perplexity in the anti-ads-in-AI-chatbots camp alongside Anthropic’s Claude, which ran a Super Bowl ad poking fun at OpenAI’s advertising plan earlier this month.
(Source: The Verge: “Perplexity joins anti-ad camp as AI companies battle over trust and revenue”)
