TL;DR
The May 2026 core update is now roughly on Day 10 and within its peak volatility window, with the most significant positional shifts expected through early June, with completion projected around June 4. On top of that, Mike King (iPullRank) published the most substantive technical piece of the week on May 29, arguing that every major AI search platform has quietly moved past simple RAG to a multi-stage agentic retrieval pipeline that most SEO teams cannot see. The CNN-Perplexity copyright suit filed on May 28 is still the biggest structural news item for the publisher and GEO worlds.
Must-cover (newsletter-worthy)
1. Core update enters peak volatility – don’t draw conclusions yet
The May update started rolling on May 21 and is now in the stretch where positions actually settle. Trackers point to the March 2026 precedent, which saw its largest swings in days 7 through 12, with YMYL verticals and aggregator-heavy domains absorbing the sharpest movement.
No completion notice has appeared on Google’s Search Status Dashboard as of this morning. Why it matters: anyone pulling conclusions from this week’s Search Console data is reading a moving table. The clean comparison is a May 20 baseline against roughly June 10 – not week-over-week deltas during the rollout. Search Engine Land – SEJ
2. Mike King: AI search has moved past RAG and most teams are optimizing for the wrong layer
Published May 29 on Search Engine Land, and shared via LinkedIn, King argues that the simple “query to retriever to top-k to answer” pipeline is dead at every major platform – AI Mode, ChatGPT Search, Perplexity, Claude, Gemini Deep Research, and Copilot have all quietly shifted to agentic RAG, which means content now has to survive multiple gatekeepers before it ever reaches a response. He published code to replicate the agentic RAG pipeline and identify where content drops out of eligibility.
Why it matters: this is a direct counter to Google’s own published guidance (and the “just do normal SEO” crowd), and it comes from the most technically credible voice in the room. He argues Google’s guidance is deliberately narrow because it covers only Google’s surface, while most AI traffic hits platforms with different indices, incentives, and retrieval logic. This is the piece to read this week. Search Engine Land – iPullRank on LinkedIn
3. CNN sues Perplexity for copyright infringement
Filed May 28 via Reuters, CNN is alleging Perplexity is distributing copyrighted content without a licensing agreement. CNN’s position – per the Reuters wire – is that licensing is the preferred path, but if Perplexity refuses to engage, legal damages are the alternative.
Why it matters: this is the AI-search citation and content-sourcing battle moving from industry frustration to active litigation, and the outcome has direct implications for how answer engines source and attribute news content going forward. Single primary source so far (Reuters wire pick-up), so treat specifics as early. Reuters via KFGO
4. Google appears to have removed the indexing lag into AI Overviews and AI Mode
Noted in Barry Schwartz’s May 25 and May 29 recaps, Google seems to have cut the delay between a page being indexed and that content becoming eligible to surface inside AI Overviews and AI Mode.
Why it matters: if confirmed, timely publishing and rapid content updates become a more meaningful GEO lever than they previously were. Flag: this is observational and community-reported, not a confirmed Google announcement – treat as a developing signal. SER recap, May 29
5. John Mueller re-states: the Quality Rater Guidelines are not a ranking manual
On May 28 (Bluesky), Mueller reiterated that the QRG describes what Google wants to reward, not how its algorithms work – prompted by SEOs over-applying YMYL category logic mid-rollout. The key Barry Schwartz nuance is worth repeating in your newsletter: the guidelines are not a ranking blueprint but an accurate description of what Google is trying to surface. Reading them as a checklist for recovery during a core update is a category error. Why it matters: this is a useful reset for an audience that tends to reach for the QRG whenever rankings move. Search Engine Roundtable
On the radar
Google Ads “Prospect Mode” is rolling out – a new targeting mode aimed at brand-unaware users, distinct from remarketing and intent-based search. SER recap, May 29
Google is testing “1st order price” labels on Shopping ads to highlight first-time-buyer discounts – a re-test of something seen roughly a year ago. SER recap, May 29
Google Discover is testing short @handle URLs for publisher profile pages. SER recap, May 29
Bing is testing narrower sitelink ad designs with smaller fonts, borders, and hover effects. SER recap, May 29
Illyes at Search Central Live Shanghai: Google strongly warns that buying or manipulating brand mentions to influence AI Overviews and AI Mode answers is treated the same way as paid links – and notably would not confirm that organic mentions even help. This directly undermines many GEO-vendor “mention-building” pitch decks. Search Engine Roundtable
Google’s first official AI search optimization guide calls GEO and AEO “still SEO” and explicitly states llms.txt gets no special treatment in Google’s systems. King’s May 29 piece argues this framing is incomplete for anyone optimizing beyond Google. SEJ
Two deadlines approaching: Google Ads short-term performance data deletion starts around June 1 – pull anything you need before then. OpenAI conversion-optimized ChatGPT ad campaigns open to select advertisers around June 5. SER recap, May 28
Google has formally appealed the search-monopoly ruling and asked the courts to pause the remedies pending appeal. SER recap, May 25
Original research spotlight
No brand-new primary study dropped in the last 24 hours. The freshest data being recirculated this week comes from King’s May 29 piece, which introduces a replicable agentic RAG pipeline and identifies five distinct gatekeepers content must survive in modern AI search – this is original technical work worth citing with full attribution to iPullRank.
Separately, a Similarweb Gen AI stats update (circulating this week) shows ChatGPT referral traffic converting at roughly 7.1%, second only to paid search, following the May 7 rollout of clickable brand links directly inside responses.
Vendor caveat: Similarweb sells AI-visibility tooling, so treat as a directional trend signal and cross-reference against your own GA4 (Google now segments AI-assistant traffic as its own channel, which helps). Similarweb
Quiet day?
Moderately quiet on net-new hard news. The dominant story is the ongoing core update rolling through peak volatility, and the most substantive new piece is King’s agentic RAG analysis.