Keeping Up With AI SEO News Without the Noise

The events that change whether AI engines can see you are published by the companies that run them. OpenAI documents its crawlers. Google documents how AI Overviews and AI Mode choose supporting links. Cloudflare announces its default bot settings. Model vendors publish knowledge cutoffs and retirement dates.

Most AI SEO news is a layer on top of those documents. Some of it is useful summary. A lot of it is vendor research with a product attached, or old advice relabelled as new.

The evidence sets a low ceiling for technique headlines. A July 2026 critical survey of 45 GEO studies (arXiv 2607.14035) concluded that "already-retrieved content can causally alter its citation or use, but no reviewed technique shows a stable, longitudinal, cross-platform causal effect on organic discoverability or downstream behavior."

So when a headline promises a new technique that gets brands into AI answers, it claims something no reviewed study has shown. Read the primary document first. Read the commentary second, if at all.

You don't need a newsletter stack. You need a short list of pages that change when the rules change. As of October 2026, these are the ones that matter.

Robots.txt changes are the ones people over-react to. If you're unsure, read whether you need to update robots.txt every time a new AI crawler appears before editing anything.

For Google specifically, the AI optimization guide (July 2026) and the AI features page are the two documents to reread when news breaks. Our guide to learning Google's AI search guidance walks through both.

Run every claim through these questions before it reaches your roadmap.

1. Who published it, and do they sell the fix? A vendor can be right. Its incentive still raises the bar. Look for funding disclosures and whether the authors sell the tactic being tested.
2. Was it measured or asserted? Operator statements are the best evidence available for their own systems, but they are assertions, not studies. Vendor studies are measurements, but check the sample, the control group and how the key metric is defined.
3. Which stage does it affect? A claim about crawler access is a different kind of claim from one about which retrieved pages get cited. More on this below.
4. Was it tested on a live engine? The widely quoted "up to 40% more visibility" figure comes from the original GEO paper (Aggarwal et al., 2023), which tested an engine the authors built, with sources already supplied in context. That says little about whether a page gets found in the first place.
5. Is it current, and is the trend real? Ahrefs found that 38% of AI Overview citations came from pages in the top 10 (March 2026). The often-quoted drop from 76% is not a measured trend, because Ahrefs changed its parsing method between the two studies.

If a claim fails two of these, file it. If it passes all five and touches access, act on it. For numbers that hold up, see AI SEO statistics you can actually trust.

AI search runs in four stages. Sorting news by stage tells you whether to act.

Parametric knowledge

This is what a model knows with no web access, frozen at its training cutoff. OpenAI lists a February 16, 2026 cutoff for the GPT-5.6 family and August 31, 2025 for the Instant path. Nothing you publish this quarter reaches those answers. See parametric knowledge for how this works, and whether a new model release changes what AI says about your brand.

Access

Whether a crawler may fetch and store your page. You control this through robots.txt per bot, your CDN or firewall, server-side rendering and snippet controls such as noindex, nosnippet and data-nosnippet. News here deserves fast action.

Retrieval

Whether your page enters the model's context for a given query. Engines rewrite queries and decide per query whether to search at all. You influence this only indirectly.

Generation

Which retrieved pages the model uses and cites. Nearly all published GEO experiments live here, and the best-designed ones found rewriting tactics ineffective or harmful.

The practical rule: access news is actionable, the rest is context. The comparison of answer engines and search engines covers the same split from the control side.

These items from 2026 came from primary sources and affected real sites. Each shows the pattern to look for.

• FAQ rich results deprecated. Google's updates log records the deprecation in May 2026 and removal of the documentation in June. Anyone still selling FAQ schema for AI answers missed it.
• Google's AI optimization guide (July 2026). It says you don't need new machine-readable files, AI text files or special markup to appear in Google's generative features, and that you don't need to write in a specific way for AI.
• Cloudflare's new defaults (September 15, 2026). Cloudflare set Training and Agent crawler categories to blocked by default on ad-displaying pages for new domains, new sites on existing accounts and existing free-tier customers. A site could lose some AI access without its owner changing anything.
• Perplexity Sonar retirement (September 27, 2026). Perplexity's Agent API makes web search opt-in, so unmigrated code can return answers with no live retrieval. If a tool or study measured Perplexity after that date, ask whether search was on.

Notice what these have in common. Each is dated, each comes from the operator, and each changes access or measurement, not prose style.

When a headline lands, work through these steps in order.

1. Find the primary document. If the article doesn't link the operator's page, search for it. If no primary source exists, the claim is commentary.
2. Classify the stage. Access, retrieval, generation or parametric. Only access changes usually need same-week action.
3. Check whether it touches your site. Read your robots.txt for each named bot. Check your CDN's bot settings. Confirm your key facts, such as pricing, appear in the HTML before JavaScript runs. The AI Crawler Checker requests a page as each AI crawler and shows which ones your firewall turns away.
4. Fix blocks, log the rest. Remove an accidental block now. Note model and ranking news in a change log with the date.
5. Re-measure per engine. Look at each platform separately, because one blended number hides which engine changed. The free checker on this site gives a snapshot for up to three buyer questions.
6. Report the change in plain terms. Tell leadership what changed, what you checked and what you fixed. Reporting AI visibility to your CEO covers how to frame it without promising outcomes.

The same claims resurface with new headlines. Primary sources contradict each of them.

• "Add schema to get cited." Google says structured data isn't required for generative AI search. Ahrefs' matched-control study found AI Overview citations fell 4.6% after pages added schema, though every page tested was already heavily cited. Existing markup does no harm. It just isn't a lever for AI answers.
• "Publish an llms.txt." Google says Search ignores it. No other engine documents reading it.
• "Allow GPTBot so ChatGPT can find you." GPTBot governs training. OAI-SearchBot governs ChatGPT search.
• "Improve your E-E-A-T score." Google says E-E-A-T isn't a specific ranking factor. There is no score.
• "Chunk your content for LLMs." Google rejects chunking as a requirement and says there is no ideal page length.
• "Ping IndexNow and AI will cite you faster." The IndexNow specification never mentions AI, and Google isn't a participating engine.
• "Track your AI rank in Google." Google's generative AI report shows impressions only. No position data exists to check such a number against.

If a news piece repeats one of these as new, the rest of it deserves the same scepticism.