Confirmed
Meta Removes Training Language From Its Commercial Content Terms
Meta's updated Commercial Terms, effective October 30, 2026, drop the word train from the clause governing how it may use content collected through commercial products. The clause now reads that Meta may use such content to provide, develop, improve, research, and maintain its Company Products, with training no longer named. A Meta spokesperson told The Verge the terms only ever applied to ads and business products. Legal observers note develop remains broad enough that scope may not have actually narrowed.
In plain termsCommercial Terms are the contract Meta's business tools run under; this is one clause inside that contract, not Meta's main privacy policy.
Why it mattersA commercial crawl clause is what many business-site operators actually read before deciding whether to block or allow a bot, so a worded change here can shift blocking decisions even if Meta's underlying practice has not moved.
Our takeThe word removed is the one word everyone was grepping for, which is exactly why a spokesperson's clarification matters more than the diff. Treat this as a terms edit to watch for enforcement, not a confirmed change in what gets trained on.
Confirmed
DeepSeek Opens Its Search And Fetch API To All Developers
DeepSeek opened its in-house search engine to developers through a new Search and Fetch API, priced at $0.75 per 1,000 calls per endpoint, with billing starting October 31, 2026. Each result carries an authority field from 0 to 4, with 3 or higher flagged as highly authoritative, plus a separate rerank relevance score from 0 to 3. DeepSeek says the index serves hundreds of millions of daily requests, and calls through its own agent tools stay free until billing starts.
In plain termsAn authority field here is DeepSeek's own 0-4 trust score for a source, separate from the 0-3 score that ranks how well a result matches the query.
Why it mattersA second large, independently-indexed search API with its own authority scoring gives agent builders and SEO tooling a cross-check against Google and Bing results, and a new place pages need to rank if the API gets adopted.
Our takeA second indexed search API with its own authority scoring is a new instrument worth comparing against Google's own signals, not a replacement for them.
Observed
Google AI Overviews Spotted Missing Their Label On Desktop
Search Engine Roundtable reported some AI Overviews appearing on Google desktop search results without the usual AI Overview label, while the same overviews still carry the label on mobile. The report, first flagged by SEO consultant Glenn Gabe on X, notes Google has had similar unlabeled AI Overview bugs before. The site asked Google for comment and had not received a response as of publication.
In plain termsThe AI Overview label is the small badge Google shows above an AI-generated answer block to mark it as machine-generated rather than a normal search result.
Why it mattersA label is often the only visible cue that an answer block on the page came from AI rather than a traditional snippet, so its disappearance changes how a searcher reads the result even if nothing about the underlying answer changed.
Our takeGoogle has shipped this exact bug before, so the likeliest read is a labeling regression rather than a deliberate rollback, but nothing here confirms either way.
Observed
Google Tests Government ID Upload For Business Profile Verification
Search Engine Roundtable reported a Google Business Profile verification option, spotted by a forum user roughly two weeks earlier, that asks owners to upload a photo of a government ID, which Google says is used only for verification and never shown publicly. The option is absent from Google's published help documentation, which lists only phone, text, email, video call and mail. The report speculates it may be tied to professional-certification verification for Local Services Ads rather than a general rollout.
In plain termsBusiness Profile is the free Google listing that shows a company's hours, location and reviews in Search and Maps; verifying it proves the owner controls the business it represents.
Why it mattersA documented verification method change affects how quickly a new or disputed local listing can get back into good standing, which is the thing local and multi-location businesses watch most closely.
Our takeAn option absent from Google's own help docs is a limited test by definition, most plausibly scoped to the professional-certification case rather than a general verification method change.
Observed
Google's John Mueller Acknowledges Old Internal Issues With Square Brackets In URLs
Responding informally on Bluesky, Google's John Mueller told former Googler Vanessa Fox that Google saw internal issues handling square brackets in URLs, the kind used in parameter syntax like filter[]=123, though the issues he recalled dated back more than ten years. Unencoded square brackets are invalid under RFC 3986, and Mueller advised consistent encoding going forward rather than relying on Google's tolerance for the characters. No Search Central post accompanies the comment.
In plain termsRFC 3986 is the technical standard defining which characters a URL is allowed to contain unencoded; square brackets are not on that list.
Why it mattersFaceted navigation and filter parameters are exactly where square-bracket syntax shows up on e-commerce sites, so a decade-old handling quirk resurfacing is a reason to check encoding on those URLs rather than assume Google treats them consistently.
Our takeAn informal social-media comment about decade-old bugs is a signal to audit your own bracket-parameter encoding, not evidence of a current Google crawling or indexing problem.
Model & AI releases
AI model launches, model API price changes and major assistant features, kept separate from the search news above. Every model story Pulse has covered is collected under LLM Models & Agents, and the monthly model comparison shows how they stack up.
Confirmed
Gemini Notebook Adds a Review Step To Slide And Video Creation
Google added a co-creation flow to Gemini Notebook's Slide Decks and Video Overviews, inserting a review stage before final rendering. A new toggle lets users read outline notes for slides, or producer notes for video, and give feedback, such as adding an executive summary, before the file generates, with a skip option for immediate generation. Google is rolling it out globally to all Notebook users over the week following its October 19, 2026 announcement.
In plain termsGemini Notebook is Google's tool for turning a set of uploaded sources into a summary, slide deck, podcast-style audio or video overview.
Why it mattersA review-before-render step changes how much a user can steer automated output without a full regeneration, which matters for anyone using Notebook to produce client-facing slides or video rather than personal drafts.
Our takeA steerable outline-before-render step is the same pattern text-generation tools already added, now reaching slide and video output, worth re-testing if Notebook output has felt un-editable before.