Live curated AI and search news feed

People were sharing screenshots of this last month and now it seems to have progressed from test to rollout - Google is using google.com/goto as a passthrough URL in Search results. It's been confirmed by a Google spokesperson who told Barry Schwartz:

> "We have a long history of deploying technical measures against evolving forms of abuse, and we regularly take steps to protect our services and users."

The idea is they want to make it harder for third parties to scrape Google SERPs. So far I am only seeing this happen when I am logged out Incognito. Just hover on a search result and look at the URL preview.

Claude Cowork desktop app has a built-in browser now. If a task requires browsing a website it'll open in a side panel and Claude will do its thing: navigating, reading, and interacting with web pages.

Anthropic says if you're using Claude Cowork on a web page that's already open, then you should stick with the Claude in Chrome extension.

> The built-in browser is for handing web tasks to Claude while you keep working: gathering research for a report, or collecting this month’s invoices from a vendor portal. Claude in Chrome is for the page you already have open, with the accounts you're already signed in to, such as updating your CRM, working through your inbox, or editing the doc in front of you.

It's rolling out now and will be enabled by default.

Google announced Gemini 3.5 Transcribe:

• Their most precise speech-to-text model yet
• Converts raw audio into accurate, polished, formatted text
• Background noise, complex jargon and "disfluency cleanup" (erasing um, like, uhh) are no problem

It's available in the Gemini API for developer use but for the rest of us it's available in the macOS Gemini app and coming soon to Chrome, so you can talk in any web field to reply, draft, or prompt Gemini in Chrome by talking.

This seems like it will result in longer, more conversational prompts/queries which may or may not lead to a website visit. When you use Gemini in Chrome, even if you are already on a web page, it does use Google Search (including citing sources) to generate responses.

I tried it out in the macOS app and the UX takes some getting used to, but it's definitely accurate!

Generative UI in AI Mode & AIO means searchers have even less incentive to click through to web pages with certain tools (now that Google can whip up mortgage calculators etc on demand).

The pages at biggest risk of irrelevancy have formula-based tools Google can easily recreate.

Time to get creative and come up with tools that do things Google can't replicate on its own yet:

• Logged-in workflows that use a customer's specific account info
• Tools integrated with an additional source (like pulling live carrier rates for shipping costs, an inventory checker that connects to your actual warehouse system, an internal product database that checks for compatibility, etc)

The article also points out that if you had a calculator page just generating traffic but no leads, you're probably better off focusing on the pages that do generate leads (if you can't make a tool that does).

Helpful concise rundown of things to check after the switch from Content API to Merchant API in GMC - attribute rules, supplemental feeds, and tracking can all be affected.

Google Scholar (Labs) has a new feature called Quick Read:

• Search your question in Google Scholar
• Click Quick Read next to a result & it opens a side panel with Answers, Approach, Consideration

Makes research go faster!

As of yesterday there's one memory across Claude Cowork & chat

Barry Schwartz dug into this as well - apparently you can still get PDFs to show up reliably in SERPs with filetype:pdf appended to your query, but it's more work than it used to be

Promotions in Google Merchant Center now have an Event section. You can create a promotion that's tied to an event like a special occasion, anniversary, holiday, and it doesn't have to be tied to a specific product or product category (but you can still add those).

Starting at the end of January 2027, the new minimum size for product images in Google Merchant Center will be 500px per side. (Google _recommends_ 1500px per side).

Don't be worried if you see an email from Google Search Console about a "new owner" on your GSC account... which is already an owner of the account. Everybody's getting them (I'm seeing them too). Not acknowledged by Google yet but it's gotta be a bug.

Really awesome resource with a really awful title (better: "How to decide which 3rd-party citations your brand should go after")

The best part is a copyable spreadsheet with a detailed what/where/why breakdown.

There's another error/more missing data in GSC, this time in the Crawl Stats report (Settings → Crawl Stats). Data is missing for August 15 & 16. This happened most recently last October. So if you notice it, it's not just your website it's a bug on Google's end. I can confirm I am seeing this too.

If you're a Mac user you can use Apple Messages with ChatGPT now. Here's the plugin documentation. It's available for all plans, but only in the ChatGPT desktop app for macOS.

They announced it on X and it includes a video example where someone is asking ChatGPT to tell them which messages they need to reply to and then using the suggested reply.

Google Search Central made this post on LinkedIn (no corresponding blog, at least not yet) about a change in how they're parsing JSON-LD structured data going forward.:

> To bring our parser up to JSON and other standards, we changed our JSON-LD extraction and are now only applying a single pass of HTML unescaping. Practically speaking, this means that double-escaped entities (like & or ✔) will no longer be unrolled.

>

> If you're using JSON-LD for structured data, be sure to update your code to standard JSON escapes or Unicode hexadecimal escapes (like \u0026).

I had to educate myself about "double-escaped entities" to get why it matters and here's the deal:

Some characters have special meaning in HTML (`&`, `<`, `>`, `"`), so they get "escaped" into a substitute called an _HTML entity_ so browsers don't misread them:

• `&` → `&`
• `<` → `<`
• `>` → `>`
• `"` → `"`
• checkmark :heavy_check_mark: → `✔`

Escaping is normal and automatic. But sometimes it happens _twice_ ( `&` becomes `&` ) and that's bad because it no longer reads right.

Google used to do multiple passes of HTML-unescaping when reading JSON-LD structured data, which let it find & fix characters that got accidentally escaped twice. Now it won't. Going forward, it does a single unescaping pass, so double-escaped entities will stay broken.

Google's recommended fix: update your code to use standard JSON escapes (no doubles!) or Unicode hex escapes like `\u0026`.

Double-escaping can happen when schema is auto-generated by a CMS plugin, GTM, or custom JS. So if you run an ecommerce or content-heavy site with auto-generated schema, check your raw source code for double-escaped entities and fix any you find. Otherwise the affected schema can be unclear (defeating the purpose of creating it!) or fail to qualify for rich results.

The August 2026 Spam Update is over already (as of 3:49am CST)! Google had said when announcing it that "it may take a few days" but it was barely even 3 days. The most recent previous spam update in June only took 2 days. Maybe the start of a new trend? Time will tell.

Google announced 3 new personalization features for Search, Discover & News.

1. New 'Preferred Sources' button. The previous iteration, when clicked, sent users to Google's source preferences tool The new version adds the site it's embedded on as a Preferred Source, then immediately brings the user back to the web page they clicked it on.
2. Discover Feed Controls (in the Google app). Click any three-dot menu in the Discover feed to adjust want you want to see more or less of, on the fly.
3. Customizable News Audio (in the Google News app, on Android). Curate the daily audio news briefing by selecting topics.

Ahrefs launched a new model for organic traffic. For now, it's opt-in only. Here's what they're saying about it:

> *Adaptive organic traffic model (opt-in)*

> We’ve launched a brand-new model for organic traffic estimates that assigns refreshed weights to SERP features, including AI Overviews, and the calculations behind those are updated monthly. Ready to test? Go to _Site Explorer → Settings → Organic traffic model_ and select _Adaptive_. what you think!

>

> We’re taking our time to make sure the new model performs as we expect, so for now it’s opt-in only.

> For some sites you’ll see only minimal changes, while for others the difference will be more pronounced.

> If you switch to the adaptive model, you’ll see updated stats across Site Explorer, except for backlink reports and the “Top organic competitors” widget.

> Historical data is also affected, though anything earlier than February 2025 uses the state of the model as it was in that month.

> You can use the new model in our API by passing the `traffic_mode=adaptive` parameter.

Just a reminder that the Content API for Google Shopping was deprecated as of yesterday, and if you haven't done so already, it needs to migrate to the Merchant API. You can check in Google Merchant Center under Settings → Data Sources and then see what it says under the Source column.

Here's Google's documentation on migrating to Merchant API, and here's a link where you can apply for extended access to Content API.

SEO Roundtable reports widespread complaints about favicons missing from Google SERPs. I've seen this for at least one client too. Always a good idea to check and make sure it's not a website issue of course. But if you're noticing a missing favicon with no cause, it could be a bug on Google's side.

Firefox has a new (Beta) feature called "Smart Window" - seems kinda similar to Google's Web Guide and they're promoting it as an AI-infused results page + assistant with more privacy and flexibility.

Google announced a new feature for Google Trends:

> We designed new interactive maps for clearer data storytelling. :world_map: Instantly spot regional trends, capture visual insights for your audience, and explore the 'where' behind every search with our new, intuitive maps.

>

> This experience is being rolled out gradually.

> Try it out at → trends.google.com/explore!

I'm not seeing it yet myself but will be keeping an eye out.

Eli Schwartz asked a question on Google Search Central's LinkedIn post about the August 2026 Spam Update, and John Mueller answered.

Q: “Why after 26 years does there need to be an update to get rid of spam instead of it just happening on its own with existing algorithms?”

A: “We have lots of systems that update continuously (most of them update automatically), but when we make bigger / broader changes to the systems, we try to announce them so that it’s easier for folks if they see a big jump in their metrics.

That said, it’s sometimes hard to find a good balance between giving some transparency & causing unnecessary worries.”

Google announced the August 2026 Spam Update within the past hour. They haven't done back-to-back Spam Updates since March/June of 2024. Nothing special about the announcement, just the usual message:

> This is a normal spam update, and it will roll out for all languages and locations. The rollout may take a few days to complete. More information about spam updates can be found at: developers.google.com/search/docs/appearance/spam-updates

> We'll update our ranking release history page when the rollout is complete.

You can make a copy of a Gemini Notebook now! I hope they extend this capability to NotebookLM.

Google confirmed there's a logging error in the new Generative AI in Search report in GSC.

Starting on August 13, impressions fall sharply. Looks like an annotation has automatically been added to the report, which is nice. Google says "This issue affects data logging only and is ongoing."

YouTube is changing how they count views starting August 24:

• A view is counted as soon as the video starts - no minimum watch time
• Engaged views will be a separate metric in YouTube Analytics (advanced mode)

So a view will be more like an impression. View numbers will go up, but mean less.

Google Research published a study effectively saying:

• Even when LLMs have all the facts they need to answer a question, pretty often they still get it wrong
• It's not a knowledge problem, it's a recall problem
• LLMs are way better at subject-object facts, and when you flip the question to ask about the object first (like on Jeopardy) they fail more

> Our results suggest that recall is tightly coupled to the conditions under which a fact was learned. When the query diverges from the training-time context, phrasing, or ordering in which the fact was encountered, recall becomes harder.

> the fact appears to be encoded, and even recognizable, but difficult to recall when the query direction departs from how the fact was encountered during training. The reversal curse is a recall problem.

*Clarity, directness → Creativity, variety*

There are no official AEO implications drawn in this research but it seems smart to stick to basic "What is A?" "A is B" type phrasing for in web content to make it easier for machines to get it right.

Apple Maps now sells ads - not a stretch to predict it will impact organic visibility for businesses on Apple Maps since Local Ads in Google Maps pushed organic listings lower on the screen

SERP Alert shared on LinkedIn that they're seeing a dip in organic in GMC analytics around 8/11 - I checked a handful of the GMC accounts we monitor and am seeing a similar pattern. No Google confirmation or comment yet but good to be aware of.

Google Sheets has a new feature: Sheets canvas. Turn your spreadsheets into "custom, interactive mini-apps" by prompting to visually organize, edit & navigate data in whatever layout you want. Rolling out starting today! Try it out when you see it: Look for the "Create canvas" option in the Gemini side panel in Sheets.

Google Business Profiles have a new option: Report owner response. Users can now tell Google if they feel a business owner's response to a review is problematic in some way.

Oh hey there's another Gemini model - but only for Enterprise/AI Pro/Ultra.

> Gemini 3.7 Flash, our most intelligent workhorse model yet for coding and agents.

Good at:

• Debugging
• Issue resolution
• First-pass code accuracy
• Layouts & apps in fewer prompts
• High design adherence to reference input
• Reasoning & accuracy
• Processing complex documents

MS Clarity added a new AI Scrape-to-Referral Ratio card in the Bot Analytics dashboard. The idea is you can see which (if any) AI-referred sessions include scrolling, conversion & other engagement signals (or if they just scrape and bounce). That way you can determine which AI traffic sources are worth something.

Gemini announced the next apps it will connect to:

• Otter.ai, Wix, Granola
• Ticketmaster, Fever, GetYourGuide, Localiza
• iHeartRadio, Pandora
• Angi, Thumbtack, Zocdoc

Rolling out "over the next few weeks" - this documentation has the full list of connected apps.

Google Sheets now does a better job importing tables & pivot tables from Excel

I just love seeing "spills the tea" and "forsook" in the same headline :pinched_fingers:

Google Merchant Center performance reporting is getting 4 updates starting August 24:

• YouTube affiliate traffic will be segmented out - no longer lumped in with Organic (you might see organic decline)
• YouTube organic traffic definitions will change so they match established YT reporting standards (again, might cause a drop in organic when this takes effect) - when it drops, this one is gonna apply to historical data going back to July 1 of this year
• Product-level reporting for Google Ads will expand to include all ad channels/formats (expect an increase - here's documentation with more info about the changes)
• There will be new ways to segment performance data (this one is not launching on the 24th, but sometime in the future)

Anthropic updated their "How Claude marks AI-generated content" support page. They're making it easier for anyone to tell when content (including plain text) was created by Claude:

> 1. Embedded watermarks in text

> When a supported Claude model generates text, it weaves an imperceptible watermark directly into the text itself. You won’t see it, and it doesn’t change the meaning, quality, or readability of Claude’s response.

>

> Because the watermark is part of the text, it will travel with the text when it’s copied and pasted elsewhere, and may persist through some editing. Watermarking will be applied at the model level, which means it will be present no matter which Claude product or surface the text comes from.

>

> 2. Signed provenance metadata

> When Claude generates a supported file type, such as a .svg, .png, or .jpg, it will attach signed provenance metadata. This metadata follows the Coalition for Content Provenance and Authenticity (C2PA) open standard, which is used across the industry to record information about content provenance. If a signed metadata label is present, it signals that a file was processed by Claude and lets you detect whether the file has been tampered with.

>

> Detecting Claude’s marks

> We’re also working to enable users and other third parties to detect Claude’s embedded watermarks and provenance metadata. Detection checks whether a piece of text or a file carries a supported Claude mark. If a supported mark is found, it indicates that the content may have been processed by Claude.

Technical documentation forthcoming.

Ask Advisor in Google Analytics & Google Ads has new agentic capabilities now:

• AIO at top of homepage with summary of important updates since you last looked
• Click on any data card to start an analysis of it with Ask Advisor
• More AI-powered insights on Google Ads homepage with a prompt box if you want custom insights
• New dashboards (in Ads now, coming soon to GA) let you create data visualizations with prompts, and it'll generate a summary explaining the data too
• New benchmarking against anonymized averages from similar businesses

Cloudflare has been busy. They made their own browser called Kitesurf, and it's designed for AI agents.

Here's a free Common Crawl visibility checker for full domains or individual URLs. Common Crawl also published an AI Visibility Audit guide.

Cloudflare released an AEO Visibility Dashboard with citation rate, mention rate, prominence (quality of coverage), and share of voice.

Plus/Pro users of ChatGPT get an updated version of GPT-5.6 Sol now (more reliable facts, more focused answers, and a slider to control how much thought goes into the response), and GPT-5.6 Luna is now the default model for Free/Go users. Starting next week they get unlimited text chats & a new "Think" button for harder questions. No changes to Work/Codex with this release, just Chat.

Ask Maps got an update:

• Agentic capabilities for multi-step tasks like ordering food and discovering local events w/links to tickets
• (Optional) Personal Intelligence integration so Ask Maps can access your Gmail and Calendar when doing tasks
• Past conversation memory (if history settings are on) so you can pick up where you left off
• New transit widget for real-time updates on bus, train, subway, or ferry rides
• Conversational contributions making it easier to suggest edits to places

SparkToro launched V3:

• Fresh UI, design, branding & main nav
• Upgrade to the new 'My Projects' home base
• New charts, graphs & data visualizations
• Upgrades to various research report elements (e.g. more graphing options, additional data)
• 'Take Action' is now called 'Workspace'
• New Help Hub with more detailed support & product demo videos
• Documentation for new(ish) API & MCP server options

Cloudflare Agents is a thing now.

> Cloudflare Agents brings all of your deployed agent sessions into a single experience, surfacing key information and insights into how your agents perform at scale.

> The Cloudflare dashboard now has a dedicated Agents view that lists observed agents and their traces alongside runs, sessions, instances, and reported token usage.

Here's the agent tracing documentation.

Reddit's 2026-Q2 letter to shareholders and statements from CEO Steve Huffman's earnings call reveal they want to be a destination for Search and decision-making, not just show up in Google Search/AI summaries.

To that end, they plan to:

• Convert more web traffic into Reddit app users by pushing the app on visitors (mostly mobile, but they've launched a desktop login upsell campaign too)
• Make the home feed into an algorithmic recommendation engine, including content from subreddits users aren't following (using machine learning to determine what content users would be interested in)
• Expand their internal search feature (Reddit Answers) to keep people on the platform
• Invest in video - they recently added video replies for comments and say they want to reimagine how users connect, share & consume content through a "modernized video experience"
• Strengthen bot-blocking and ML-driven proactive moderation to fight spam even better and "keep Reddit human and safe"

SEJ article with "Reddit as a search destination" analysis of call

SEJ article with "Reddit making better use of their evergreen content" analysis of call

MS Clarity added branded/non-branded labels for grounding queries in the AI Citations dashboard

More Gemini capability added to Google Docs:

• Comment workflows enable Gemini to read, summarize, and act on comments in Docs
• Create & edit images, diagrams & infographics with Gemini in Docs