The stories, shifts and signals from this week in AI Search, and what they could mean for your business.

    Back to Blog
    Connor Houghton··4 min read

    This Week in AI Search (w/c 13/07/26)

    Share this article

    TL;DR

    The week of 13 July 2026 was the week Google stopped sending clicks and started writing answers. Gemini 3.5 Flash now writes the primary answer for every Google search, a rollout completed 10 July, with ranked links pushed below the AI response and click-through rates on top pages down 58%. The same week, Hachette, Cengage, Elsevier and author Scott Turow filed a class action alleging Gemini trained on copyrighted books without permission, claiming Google stripped copyright metadata to hide the sourcing. Google DeepMind chief Demis Hassabis proposed an industry-funded body to test frontier models before release, winning rare public praise from Sam Altman, Satya Nadella and Elon Musk. Google also delayed Gemini 3.5 Pro a second time over weak coding performance, and Perplexity shipped SPACE, a sandboxed runtime that cut agent sandbox creation time from 185 to 60 milliseconds.

    Google made AI answers the default search result this week, ending 27 years of ranked blue links. Publishers sued Google the same week over the training data behind that shift. Google DeepMind's chief published a regulation framework rivals openly praised.

    What Happened

    1. Google Makes AI Answers the Default Search Result

    Gemini 3.5 Flash now writes the primary answer for every Google search, a rollout completed 10 July (TechTimes, 2026). Ranked links remain but sit below the AI response. Click-through rates on top pages have fallen 58% since (TechTimes, 2026).

    2. Publishers Sue Google Over Gemini Training

    Hachette, Cengage, Elsevier and author Scott Turow filed a class action against Google in New York, alleging Gemini trained on copyrighted books without permission (TechCrunch, 2026). The suit claims Google stripped copyright metadata to hide the sourcing, citing an internal document flagging fines of "$10Bs-$100Bs" (TechCrunch, 2026).

    3. Hassabis Publishes an AI Regulation Framework

    Google DeepMind chief Demis Hassabis proposed an industry-funded body to test frontier models before release (Axios, 2026). Sam Altman, Satya Nadella and Elon Musk praised the plan. It marks the first time Google, OpenAI and Anthropic's leaders have converged on one regulatory approach (Axios, 2026).

    4. Gemini 3.5 Pro Misses a Second Deadline

    Google delayed Gemini 3.5 Pro again, citing coding performance below internal targets (9to5Google, 2026). Engineers retrained the model on updated coding data in June, but results disappointed.

    5. Perplexity Ships SPACE Runtime for AI Agents

    Perplexity launched SPACE, a sandboxed runtime now powering every session of its Computer agent (TestingCatalog, 2026). Sandbox creation time dropped from 185 milliseconds to 60 milliseconds in testing (TestingCatalog, 2026).

    Why It Matters

    Control shifted decisively toward Google this week. Its AI answer sits above every result by default, and defaults decide what most users read first (TechTimes, 2026). The company that indexed the web now writes the summary instead of linking to it.

    Publishers footing that bill are fighting back. Hachette, Cengage and Elsevier joined a lawsuit accusing Google of training Gemini on unlicensed text, the same week Google's AI began answering more queries without linking out (TechCrunch, 2026). Click-throughs on top pages have dropped by more than half.

    Regulation looks closer than it has all year. Hassabis, Altman and Amodei now agree frontier models need outside testing before release, a rare consensus among rivals (Axios, 2026). The same companies now write the answers users read and the rules meant to govern them. To see where your business stands inside AI search today, run your free AI Discoverability Score.

    Who Wins / Who Loses

    Winners

    Google: Locked in AI answers as the default for billions of searches, cementing Gemini's reach before rivals catch up.

    Perplexity: SPACE cuts agent sandbox latency up to fivefold, strengthening Computer against Google and OpenAI (TestingCatalog, 2026).

    Demis Hassabis: His regulation framework won rare public praise from Altman, Nadella and Musk (Axios, 2026).

    Losers

    Publishers: Hachette, Cengage and Elsevier fight Google in court while losing traffic to the AI they accuse of training on their books (TechCrunch, 2026).

    Google's Gemini coding team: A second missed deadline leaves the flagship model behind schedule (9to5Google, 2026).

    Independent websites: Click-throughs on top pages fell 58% as AI answers replaced links (TechTimes, 2026).

    The One Thing to Remember

    Google stopped sending clicks and started writing answers. That shift, completed this week, changes who controls what users see across the open web. Publishers are suing over the training data behind it, regulators are circling the frontier labs, and rivals are racing to build their own version of it. The search results page as it existed for 27 years is gone.

    References

    • 9to5Google (2026) Gemini 3.5 Pro delays due to coding performance, upgraded Flash model in testing. Available at: 9to5google.com (Accessed: 20 July 2026).
    • Axios (2026) Behind the Curtain: AI godfathers converge on regulations. Available at: axios.com (Accessed: 20 July 2026).
    • TechCrunch (2026) Google faces another AI training lawsuit from major publishers. Available at: techcrunch.com (Accessed: 20 July 2026).
    • TechTimes (2026) Google Replaced Its Default Search With AI: How to Get Blue Links Back. Available at: techtimes.com (Accessed: 20 July 2026).
    • TestingCatalog (2026) Perplexity launches SPACE runtime for AI agent tasks. Available at: testingcatalog.com (Accessed: 20 July 2026).

    Keep Reading

    Related Articles

    Continue reading on similar topics

    Start With Clarity, Not a Contract

    See where you show up when buyers ask AI who to hire, and who shows up instead of you.

    Book a Free Growth Assessment

    We show you the gap, you decide what to do with it.