Google tightens bot defenses with /goto redirects, adds ARD audit to Lighthouse, and extends spam update pressure to AI-generated content

Google prepares /goto redirects for wider rollout against automated search traffic

com/goto`, toward broader deployment. The mechanism began testing in July 2026 and is slated for wider deployment to logged-out users within the next month. Its purpose is to add another hurdle for bots, including rank-tracking tools and AI platforms that issue automated queries against Google Search.

For practitioners, the practical consequence is that server logs and rank-tracking dashboards may start showing new redirect paths, and any tooling that follows raw result URLs without following redirects could begin to drop or misreport data. The change is positioned as part of Google's longer-running effort to distinguish automated agents from human searchers and to keep result-scraping infrastructure from distorting analytics.

Lighthouse 13.5 ships an experimental AI Agent Resource Discovery audit

Google Lighthouse 13.5 added an experimental "Agentic Resource Discovery" (ARD) audit, giving the SEO tooling community its first mainstream way to test whether sites expose catalogs that AI agents can consume. The audit checks ARD catalogs, an early-stage proposal for how agents discover resources on the web. A practical caveat flagged by one technical analysis: Lighthouse's checks lag the newest ARD proposal, so a site can fail the audit or report Not Applicable even after implementing the latest draft. Practitioners should treat results as directional, not pass/fail, and should not reorganize information architecture around a Not Applicable outcome.

August 2026 spam update pressures scaled AI publishing

Google's August 2026 spam update rolled out globally between August 18 and 21, 2026, and was the third confirmed Google spam update of the year. While Google did not officially frame the update as targeting AI content alone, broader enforcement against scaled, low-value publishing is hitting sites that use generative AI primarily to manufacture search traffic rather than help readers. For sites with large templated footprints, the operational task list is to audit thin pages, consolidate near-duplicates, and confirm that each URL has a clear reader-facing purpose independent of ranking.

Publishers leaning on automation should treat this update as the third signal in a year that scaled content without proportionate value is the enforcement target.

What practitioners should verify in the next two weeks

Three checks are worth running while the /goto rollout completes and the ARD audit stabilizes. First, compare a small sample of logged-out searches in your rank tracker against manual searches and note any increase in redirect hops or empty result rows. 5 against representative URLs to baseline the ARD audit, but expect Not Applicable on sites that follow the most recent draft.

Third, for sites with templated or AI-assisted content, prioritize an inventory of pages most likely to read as scaled and low-value before any further expansion. txt Checker](/seo/ai-bot-robots-checker/) covers the agents most teams forget to allow or block, and the GEO Brand Question Generator helps stress-test whether your entity information answers the queries AI assistants already surface.