AI, Autocomplete, AI Search. In That Order.

AD

Help make better ads

Did you recently see an ad for Roku Ads Manager in a newsletter? We’re running a short brand lift survey to understand what’s actually breaking through (and what’s not).

It takes about 20 seconds, the questions are super easy, and your feedback directly helps us improve how we show up in the newsletters you read and love.

If you’ve got a few moments, we’d really appreciate your insight.

Take the survey.

SEO + GEO

Core Roars Back – Interesting Cases and Findings From Google’s May 2026 Broad Core Update

Glenn Gabe reports that Google’s May 2026 broad core update hit faster and harder than March, starting May 21 and finishing June 2 after an 11-day rollout. He saw impact about 48 hours in, with sharp gains and losses across countries, heavy volatility in YMYL categories such as health, finance, and gambling, and two late-rollout tremors that also affected Google Discover. Gabe’s AI search checks were more mixed: sites using risky AI search tactics were not uniformly hit, but January’s “commodity content” losers did not recover, and one site scaling low-quality content plus AI translations dropped hard.

YouTube Is Outranking Your Entire SEO Team on 1,723 Keywords

Foundation Team reports that YouTube is winning high-intent B2B SaaS SERPs that many teams still treat as side-channel territory. Foundation analyzed 8,566 keywords across 130+ SaaS categories and found YouTube in Google’s top 10 for 80.2% of “demo” queries, 22.5% of “best” queries, and 20% of “vs” queries, with YouTube outranking three or more vendors on 1,723 keywords representing 513,000 monthly searches. The sharper point: YouTube over-indexes where buyers are close to action, including demo, trial, review, setup, and comparison searches.

How to get indexed by ChatGPT [2026]

Amy Rigby explains that getting indexed by ChatGPT is not the same as showing up in a ChatGPT answer. OpenAI’s bots may fetch a page, store parts of it in an indexed or cached layer, and still never cite it in a user-facing answer. Her advice starts with allowing OAI-SearchBot, then moves through Bing sitemap submission, IndexNow, fresh lastmod signals, internal links from pages bots already revisit, and raw HTML content that AI crawlers can read without JavaScript.

My Take: This is where GEO gets boring in the best possible way. Before you chase prompts, mentions, or fancy dashboards, make sure the machines can fetch the page and see the facts.

Why 62% of AI citations don’t lead to brand mentions [Study]

Margarita Loktionova reports that AI citation visibility and brand visibility are now two different jobs. Semrush partnered with Kevin Indig to analyze 3,981 domain appearances across 115 prompts, 14 countries, and four AI search engines: ChatGPT, Google AI Overviews, Gemini, and Google AI Mode.

• Most citations hide the brand: 61.7% of AI citations were “ghost citations,” where the source link appeared but the brand name did not. Only 13.2% of appearances were both cited and mentioned.

• Engines behave differently: Gemini mentioned brands in 83.7% of appearances but cited them only 21.4% of the time. ChatGPT did the reverse, with an 87% citation rate and a 20.7% mention rate.

• Markets change the outcome: India and Sweden had brand mentions in 50% of AI answers, while Italy, Brazil, and the Netherlands landed between 18% and 22% despite high citation rates.

• Prompt shape matters: Short, conversational queries produced nearly 100% brand mention rates. Longer structured prompts fell to 2% to 3%.

• Intent changes visibility: Informational queries had an 89.3% citation rate but only an 18% mention rate. Comparative queries produced 2.4x more brand mentions than informational queries.

My Take: This is why I would separate AI search reporting into two dashboards: evidence and memory. Citations show which pages models trust enough to use as evidence. Mentions show whether the model already knows the brand well enough to say its name. SEO can help with the first, but the second needs positioning, PR, reviews, comparisons, and repeated brand facts across the web.

Google Chrome tests another massive change to Search that puts AI in front

Nickolas Diaz reports that Google is testing a Chrome Canary flag that would send address-bar searches directly into AI Mode instead of the standard Google results page. WindowsReport found the flag labeled “Fulfill Searchbox queries in AI Mode,” but Google’s commit says there are “no current plans” to ship the change and that it is “just for exploration.” If it ever moves beyond testing, Chrome would turn the browser’s default search box into an AI answer surface where links appear only as supporting sources.

My Take: This is the search behavior change SEOs should watch more than any single AI Overview tweak. If the browser pushes users into AI Mode before they ever see a classic SERP, ranking position becomes only one part of visibility. The new job is making pages easy for AI systems to retrieve, trust, and cite when the answer interface becomes the default front door.

WTF is NLWeb? — Whiteboard Friday

Crystal Carter explains NLWeb, Microsoft’s open-source Natural Language Web project that uses the ASK protocol to let people and agents query a website in plain language. The system can pull from schema, headings, title tags, HTML lists, tables, product feeds, RSS feeds, XML sitemaps, and CMS data, then ground answers in the site’s own information. She also notes that R.V. Guha, the creator of Schema.org and RSS, created NLWeb, which makes the project worth taking seriously even if adoption is still early.

My Take: NLWeb is a reminder that all the “boring” SEO work still matters. Clean structure, schema, headings, feeds, and well-organized content are becoming the connection layer between websites and agents. I would not rush every client into implementation tomorrow, but I would use NLWeb as a checklist for whether a site is ready to be queried by machines instead of just crawled by bots.

How AI prompt patterns vary by industry and shape search visibility

Jason Tabeling explains that prompts are becoming the new unit of search because AI systems retrieve different sources depending on how users frame the task. Healthcare prompts often include symptoms, age, medication, and risk constraints, while B2B prompts ask for comparison tables, implementation timelines, hidden costs, and ROI. Ecommerce prompts blend “best,” “cheap,” reviews, use cases, and budget limits, while citing research that nearly 45% of LLM follow-up nudges are budget- or deal-related.

My Take: This is why keyword research alone is starting to feel incomplete. It’s all about topic research and looking at a higher level. A topical map now needs a prompt layer to find the topic gaps. You need to know what buyers ask first, what they ask next, and what facts they need. This is why we have the AI search gaps feature to find the topic gaps after doing Google and YouTube autocomplete.

How to estimate the traffic impact of SEO fixes

Adam Heitzman shows how to forecast SEO fixes by defining the scope, calculating current traffic exposure in Google Search Console, estimating lift from past projects, and presenting conservative, expected, and aggressive scenarios. His warning is useful: the same ranking gain can produce different traffic depending on SERP features, AI Overviews, and query type. He cites SparkToro data showing 68% of U.S. Google searches in early 2026 ended without a click, plus Seer Interactive research showing organic CTR for informational queries with AI Overviews fell 61% between mid-2024 and late 2025 before rebounding to 2.4% by February 2026.

My Take: This is how SEO earns roadmap trust. Clients and founders do not need perfect predictions. They need to know which fix is worth developer time, which one can wait, and which one only sounds important because it is technically interesting.

What co-mentions reveal about the AI recommendation gap

Maryanna Franco reports on a study of 12 athleisure and activewear brands across ChatGPT, Gemini, Perplexity, Claude, and Google AI Overviews, using 14,140 API runs over seven days in the UK. The study found that Knowledge Graph strength did not guarantee recommendation visibility: Nike appeared in 71% of athleisure recommendation prompts, while New Balance and Reebok appeared in 0%, even though all three shared the same Google Knowledge Graph description as “Footwear company.” Recommendation prompts relied heavily on outside validation, with third-party sources accounting for 82% to 100% of citations across the five systems.

Schema.org Adds Usage Statistics For Schema Types

Schema.org has added usage statistics to individual schema type pages, giving SEOs a public way to compare adoption across the web. The dataset updates monthly, aggregates usage at the domain level, and groups popularity into buckets to reduce day-to-day noise. Schwartz notes that author schema appears on over 10 million domains, while event schema appears on under 1 million domains, which makes the gap between common and neglected structured data easier to spot.

My Take: This is useful because schema work has always suffered from too much opinion and not enough market context. If a markup type is common, it becomes table stakes. If it is rare but relevant, it may be a chance to make a page easier for search systems and AI systems to parse. I would use this dataset as a prioritization layer, not a rulebook.

Apple Updates Applebot Docs To Include Siri AI & AI Features

Barry Schwartz reports that Apple updated its Applebot documentation on June 8, 2026 to explain how crawled data may support AI answers in Apple products and services, including Siri and Search. Apple says publishers can use nosnippet to opt specific content out of broad world knowledge answers, while Applebot-Extended remains the control for opting out of foundation model training. The update also adds guidance for X-Robots-Tag directives, paywalled content via isAccessibleForFree, and crawl behavior, including Applebot not honoring crawl-delay.

Google Updates Its Hiring An SEO Doc With SEO Tools & Generative AI Optimization

Google updated its Do you need an SEO? document with new warnings about SEO tools, audits, and generative AI optimization. Google now tells site owners to check whether an SEO’s AEO or GEO advice aligns with Google’s official guidance, to grant Search Console read access only during an audit, and to avoid anyone promising first-place rankings. He also notes that the page is about 30% shorter than the archived version, with the new third-party tool warning highlighted in red.

My Take: No comment 😅 

Google On Third-Party SEO Tools, Services, and Advice

Google published a new third-party SEO tools and advice guide for evaluating external recommendations. Google says tool vendors and SEO services may help with sitemap generation, indexing directives, “SEO-optimized” content, ranking advice, and AEO or GEO claims, but it does not evaluate or endorse third-party tools. The warning is direct: third-party tools do not have Google’s internal ranking data, cannot guarantee performance, and should be checked against official Google Search guidance.

My Take: No comment again 🥱 

Claude visibility may depend heavily on Brave Search rankings, new data suggests

Danny Goodwin reports that Claude may be more tied to Brave Search rankings than other AI answer engines, based on Jonathan Clark’s takeaways from a Zero Click by Profound session. Clark said Claude used web search in 36.6% of prompts, versus about 90% for ChatGPT, and was most likely to search on freshness, ranking, location, and comparison prompts. Claude citations matched ChatGPT only 8% of the time, but overlapped with Google rankings 64%, which makes Brave and Google rank tracking more useful for Claude visibility tests.

SEO + GEO Ripples

• Bing Gives Searchers A Way To Disable AI Copilot Answers: Bing is testing a Chrome and Edge extension that lets users turn Copilot-style AI answers on or off in search results. Users can also add -ai to a query to suppress AI responses, which turns user choice into a visible search product feature instead of a settings-page afterthought.

• Google Search Console Link Report Finally Fixed & Updated: Google has fixed the Google Search Console links report after weeks of missing or sharply reduced link counts.

• Google Is Adding Business Profile Tools To The Gemini App: Matt G. Southern reports that Google is adding a direct Google Business Profile connection and Business notebooks to the Gemini app, rolling out globally this month outside the EEA and UK. Once connected, Gemini can answer performance questions, draft review replies, suggest profile updates, and carry business context across notebook sessions.

• US Publishers Demand Common Crawl Stop Scraping Their Content: Digital Content Next sent Common Crawl a cease and desist letter demanding that it stop collecting protected publisher content and remove member material already in its datasets. Blocking CCBot only stops future crawls, while older Common Crawl archives can still sit inside AI training pipelines.

• More News Sites Default To Blocking AI Crawlers: Reuters and Time now block AI bots by default and approve crawlers only through allowlists. Reuters says the setup has not cost traffic, while People Inc. saw blocked user agents jump from about 2,100 to more than 30,000 after moving from blocklists to allowlists.

• Google Says Hyphenated Domain Names Are Okay For SEO: Roger Montti reports that Google’s John Mueller said hyphenated domains are “ok” for SEO, pushing back on the old assumption that dashes carry a hidden ranking penalty. The potential issue is still branding and trust: domains with hyphens can be harder to type, easier to misread, and more likely to look spammy if the rest of the site does not build confidence.

AI

Claude Fable 5 and Claude Mythos 5

Anthropic releases Claude Fable 5, its first Mythos-class model for general use, with Claude Mythos 5 reserved for trusted cyberdefenders, infrastructure providers, and future biology partners. Fable 5 falls back to Claude Opus 4.8 for flagged cybersecurity, biology, chemistry, and distillation requests, and Anthropic says those safeguards trigger in less than 5% of sessions on average. The release pairs frontier-model access with a stricter operating model: 30-day retention on Mythos-class traffic, $10 per million input tokens, $50 per million output tokens, and staged subscription access that changes after June 22 - oh, but wait…

Statement on the US government directive to suspend access to Fable 5 and Mythos 5

Just as quick, Fable is gone - Anthropic says the US government issued an export control directive on June 12 to suspend all Fable 5 and Mythos 5 access for foreign nationals, forcing Anthropic to disable both models for all customers to comply. The company says the order cited national security authorities but gave no specific details, and Anthropic believes it stems from a narrow, non-universal jailbreak that asks the model to inspect a codebase and fix flaws. Anthropic says its red teams found no universal jailbreak, that disclosed issues show no Mythos-specific uplift, and that all other Anthropic models remain available.

WebMCP Can Be Used To Hijack AI Agents, Chrome Warns

Roger Montti reports that Chrome is warning developers about WebMCP prompt injection risks, especially malicious manifests and contaminated tool outputs. Browser-based agents may operate inside authenticated sessions, where a poisoned tool description, user comment, review, or forum post can steer an agent toward unsafe actions. Chrome’s guidance says model safety layers cannot guarantee protection, so developers need deterministic controls such as token limits, origin restrictions, untrusted content handling, and user confirmation before state-changing actions.

When AI builds itself

AI is already accelerating AI development inside Anthropic, even before full recursive self-improvement exists. As of May 2026, more than 80% of merged code in Anthropic’s codebase was authored by Claude, and the typical engineer was merging 8x as much code per day in Q2 2026 as in 2024, though Anthropic says lines of code overstate true productivity. For research velocity, Claude moved from about a 3x speedup on a fixed optimization task in May 2025 to about 52x by April 2026, while humans still hold the advantage in choosing which problems matter.

AI Ripples

• Fluid, natural voice translation with Google’s Gemini 3.5 Live Translate: Gemini 3.5 Live Translate can detect 70+ languages and generate near real-time translated speech that preserves intonation, pacing, and pitch. The model is rolling out through the Gemini Live API, Google AI Studio, Google Translate on Android and iOS, and a private preview for Google Meet, where Google says it will support 2,000+ language combinations.

SOCIAL MEDIA

• YouTube expands in-app messaging: Andrew Hutchinson reports that YouTube is expanding YouTube Chat to the U.S. and other global locations after earlier testing in Ireland and Poland. Eligible users can invite friends into private in-app conversations, share videos, react in real time, and keep more YouTube sharing behavior inside the platform instead of pushing it to outside messaging apps.

WAYS WE CAN WORK TOGETHER

Floyi - Build Topical Authority that wins in Google and AI Search. Don’t just plan your content strategy - make it unstoppable.

TopicalMap.com Service - Let us do the heavy lifting. We handle the research, structure, and strategy. You get a custom topical map designed to boost authority and dominate your niche and industry.

Topical Maps Unlocked 2.0 - Unlock the blueprint to ranking success. Master the art of structuring content that search engines (and your audience) love - and watch your rankings soar.

AD

For a limited time, we’re giving away a whole folder of resources for getting started with TikTok Shop, including an internal case study on how went from $0 to $90K/month in 90 days. 

Claim yours:

Get Your Free Starter Kit