Confirmed
Meta Delays Its Updated Crawling Terms From October 30 To November 20
Meta pushed back the effective date of its updated Commercial Terms and Self-Serve Ad Terms from October 30 to November 20, 2026, after complaints about a clause permitting automated software to create test credentials and crawl pages tied to a business's products or services. A Meta spokesperson said the extension gives partners more time to understand what is changing. The crawling permission itself, which does not acknowledge robots.txt, is unchanged in both agreements.
In plain termsCommercial Terms are the contract a business agrees to when it uses Meta's ad and business tools, and this one clause inside it covers when Meta's own bots can crawl a business's pages.
Why it mattersA business-site operator relying on robots.txt to block Meta's crawlers gets three more weeks before the clause takes effect, not a change to what the clause actually permits.
Our takeDelaying the date while leaving the permission untouched reads as a communications fix, not a policy reconsideration. Watch whether the delay buys Meta time to add a robots.txt carve-out, or just time to explain why it isn't adding one.
Confirmed
Shopify's Rollouts Tool Adds Scheduled And Tested Catalog Changes
Shopify's October 23, 2026 changelog update lets merchants use Rollouts to activate, deactivate, or switch between catalogs at a set date and time, run experiments comparing catalogs on a live storefront, and schedule catalog changes to launch alongside theme and checkout updates in one coordinated release. The feature targets availability and pricing changes tied to promotions, seasonal transitions, or market-specific rollouts.
In plain termsA catalog here is the specific set of products, prices, and availability a storefront shows; Rollouts is Shopify's tool for turning features on or off for some stores without a full redeploy.
Why it mattersCatalog changes tied to pricing or availability can now be scheduled and tested with the same precision merchants already apply to theme and checkout launches, cutting the manual-toggle risk around promotional go-live moments.
Our takeA scheduler for catalog swaps treats pricing and availability changes with the same release discipline as a code deploy, which matters most for merchants who have been doing this by hand before a sale.
Observed
Google Discover Tests Disabling Clicks On AI Summary Sources
Marketer Damien Andell spotted the test on X on October 23, 2026 and named the internal flag disable_aio_sources_button_tap. Search Engine Roundtable reported on it October 26, 2026, writing that Google Discover's AI summary cards may no longer open the mini-feed of linked articles when a reader taps the sources button. The publication could not reproduce the behavior, and Google has not confirmed the test's scope or whether it will ship.
In plain termsDiscover is the feed of articles Google shows mobile users outside of search; the sources button is what lets someone see and click through to the original articles behind an AI summary there.
Why it mattersIf the sources tap stops opening the linked-articles feed, a Discover AI summary could stop sending any click to the pages it cites, so publishers watching Discover referral traffic should track this test rather than assume current click behavior holds.
Our takeAn internal flag name leaking before any announcement follows the same pattern as past Google AI Overview source changes: it ships quietly to a slice of users well before anyone outside gets a confirmation. Treat any Discover click-through drop over the next few weeks as possibly this test, not a ranking 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
Moonshot Launches Kimi K3.1 Pro With A 1-Million-Token Context Window
Moonshot made Kimi K3.1 Pro available on Kimi.ai, Kimi Work, Kimi Code, and its API on October 23, 2026, with full model weights to follow by November 4. The model carries a 1-million-token context window. Moonshot reports Kimi K3.1 Pro scoring 65.7% on Terminal-Bench 4.0, ahead of GPT-6 Astra's 58.2% and just behind Claude Opus 5.5's 66.4%, and says its internal hallucination rate dropped from 23.4% to 8.3% versus the prior K3 model.
In plain termsTerminal-Bench 4.0 is a benchmark that scores a model on running real command-line coding tasks, not on answering trivia, so a high score there speaks to how well a model can act as a coding agent.
Why it mattersA soon-to-be-open-weight model scoring within a point of Claude Opus 5.5 on an agentic coding benchmark gives teams building on open weights a frontier-tier option to re-test once the weights ship November 4.
Our takeA vendor's own benchmark numbers are the best case Moonshot could build for its own model, so the Terminal-Bench 4.0 score is worth re-checking against an independent run once the weights are actually public on November 4, rather than banked on today.