Episode 1 of AI for Marketers. We start with a fast news rundown, then the anchor tutorial: how to choose and evaluate an AI assistant for your own marketing work, and how to keep reading the field as models change month to month. **News this week (June 2-9, 2026)** - Google is adding an [AI-search opt-out toggle in Search Console](https://9to5google.com/2026/06/02/google-ai-mode-overviews-opt-out/), driven by the UK Competition and Markets Authority and [reportedly called a "world first"](https://www.computing.co.uk/news/2026/google-to-let-publishers-opt-out-of-ai-search-features-while-remaining-in-search-results). It [won't take effect until June 17](https://www.business-standard.com/technology/tech-news/google-test-ai-search-opt-out-publisher-zero-click-web-explained-126060400852_1.html), [excludes the Gemini app](https://9to5google.com/2026/06/02/google-ai-mode-overviews-opt-out/), and is [not a ranking signal](https://www.ghacks.net/2026/06/04/uk-regulator-orders-google-to-let-publishers-opt-out-of-ai-search-summaries/). Coverage from [TechCrunch](https://techcrunch.com/2026/06/03/publishers-will-be-able-to-opt-out-of-ai-search-thanks-to-new-regulation/). - Visibility check: Search Console now has [generative-AI performance reports](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) (impressions only, by page/country/device; no clicks, CTR, or queries), corroborated by [pikaseo](https://pikaseo.com/articles/google-search-console-ai-performance-reports-2026). Per [Goodie's 2026 AI Search Traffic Report](https://higoodie.com/blog/ai-search-traffic-report-2026/), ChatGPT's share of measurable B2B AI referrals reportedly fell from ~89% to 62.6%, Claude surged to 18.5% (a 27.2% April peak), Gemini 10.6%, Perplexity 7.3%. AI Overviews reportedly [2.5B+ monthly users](https://smallbizmarketing.agency/blog/how-ai-overviews-are-changing-google-search). - HubSpot upgraded its [Claude connector to SQL-based retrieval](https://martech.org/hubspot-adds-another-crm-connector-this-time-with-claude/) and expanded its [ChatGPT connector](https://www.cmswire.com/customer-experience/hubspot-launches-first-crm-deep-research-connector-to-chatgpt/) ([release notes](https://releasebot.io/updates/hubspot)). - [Constant Contact launched an app inside ChatGPT](https://www.prnewswire.com/news-releases/constant-contact-launches-app-in-chatgpt-to-power-ai-driven-email-marketing-302794355.html) for end-to-end email campaigns with publish-back. **Tutorial: picking and evaluating your assistant** Sort the [seven major assistants](https://openai.com/index/introducing-gpt-5-5/) by your real work, your data privacy, your existing stack, and budget (the [~$20/mo individual tier](https://claude.com/pricing) is the standard). Then read the field with your own [brand-voice test](https://www.success.com/how-to-train-ai-to-match-your-brand-voice-a-guide-to-personalized-prompting), research/grounding test, and factual probe instead of trusting [gamed leaderboards](https://blog.collinear.ai/p/gaming-the-system-goodharts-law-exemplified-in-ai-leaderboard-controversy). The biggest pitfall: [hallucination](https://www.digitalapplied.com/blog/ai-model-hallucination-rate-benchmarks-2026-study) (~3.1-19.1%, citations worst at ~12.4%). Demand sources and click every link. Check [data controls](https://help.openai.com/en/articles/7730893-data-controls-faq) before pasting anything sensitive.