AI search and answer engines changed in ways that matter to two groups: people who use them to find information, and people who publish or sell online and depend on being found. Between August and early November 2026, the dated items we could find covered new Google controls, publisher and licensing arrangements, advertising inside AI answers, and reporting tools for AI visibility.
This article collects those items with dates and sources. Almost all of them come from search-industry digests rather than from the vendors’ own pages, so we label them as reported and tell you where to confirm. We have not tried to predict where this goes next.
How we compiled this: we did not run hands-on tests. This is based on public reporting and vendor pages, checked on 11 November 2026. Many items below are second-hand reports from industry trackers, and some come from single named commentators. Treat them as leads to verify, not settled facts. This is general information, not legal or financial advice.
Key takeaways
- On 15 October 2026 Google published “More ways to connect with the web from Search”, described in a digest as adding inline links, a test section highlighting official websites, expanded news carousels and clickable creator profiles in AI Mode.
- A digest reports that Google, Microsoft, OpenAI and Anthropic will support one shared robots.txt token for AI training opt-out, with Google planning support in early 2027, and that Bing’s separate grounding controls arrive on 15 December.
- Alphabet’s third-quarter results, as reported, showed search revenue growing 12 percent year over year, down from 17 percent in the second quarter, and Google said AI Mode earns less from ads than traditional Search.
- ChatGPT Ads were reported available in 89 countries in October, and OpenAI was reported to have fixed a bug that stripped its referral tag from links around 8 to 15 October.
- Reporting on AI search is still uneven. Several claims here rest on one commentator or tracker, so check before you act on them.
Summary
The items fall into four groups.
- Google adjusting the balance between AI answers and links. New inline links and source highlights in AI Mode, alongside expanded ads in AI Mode.
- Rules for AI crawlers. Moves toward shared opt-out signals and separate controls for training and grounding.
- Money. Ads inside ChatGPT, publisher payment pilots and pay-per-use schemes for AI access to content.
- Measurement. Google’s Search Console AI reports, Microsoft Clarity’s AI visibility features and the messy state of referral data.
Where a date is a month only, that is all the source gave.
Google: AI Mode, Overviews and Search Console
Search and AI Mode changes
An October 2026 digest from Everything PR reports that on 15 October Google published “More ways to connect with the web from Search”, describing inline links, a test section highlighting official websites, expanded news carousels, image thumbnails on related-article links and clickable creator profiles in AI Mode. It says Google stated that use of Preferred Sources had grown more than sevenfold since the start of the year, without giving a baseline. The same digest reports that on 13 October Gemini 4 Argon rolled out to Google AI Pro and Ultra subscribers, which a second digest says also applies in AI Mode.
A separate digest from Position Digital lists, for August, an AI Mode prompt box appearing inside AI Overviews for some queries, and for November, AI Overviews appearing in the right-hand column where the knowledge panel used to be on many desktop brand searches. It also cites a test by Chris Long finding AI Overviews on 93 percent of tested brands. That is one commentator’s test with an unstated method, so treat it as indicative at most.
Search Console and measurement
The Position Digital digest reports that generative AI performance reports rolled out to all sites worldwide in August, that Search Console added multimodal reporting covering Lens, Circle to Search, image uploads and right-click image search in September, and that Google’s Gary Illyes said in November there was nothing to announce yet on click and query data for the AI performance report. It also reports that Marie Haynes saw clicks jump between 2 and 6.5 times from 31 October to 2 November while impressions stayed flat, which she attributed to bot traffic or a logging bug. Again, this is one commentator’s observation.
Ads and spam updates
The same digest reports that ads in AI Mode expanded to 10 more countries in October, with multiturn ads in US follow-up queries, that Google completed a September 2026 spam update, the fourth confirmed spam update of 2026, after nearly 14 days, and that an August spam update was reported to target sites mass-publishing AI-generated content. It also reports Google adding UTM parameters to outgoing Gemini links for better referral attribution.
What it means: if you publish content, the practical changes are new reporting, more links in AI Mode and an unsettled picture of referral traffic. See our guide on how to choose an AI tool for a framework to judge these tools.
ChatGPT search and ads
Everything PR reports, for October 2026: on 1 October a “Try on” button appeared on clothing and accessory listings in ChatGPT shopping, and on 2 October Finances expanded to Free and Go users in the US. The Position Digital digest adds that ChatGPT Ads were available in 89 countries with a wider desktop ad card and multi-image ads in open beta, and that LiveRamp lets advertisers activate first-party audiences in ChatGPT Ads. A timeline of OpenAI release notes lists a visual ad format and measurement tools for ChatGPT ads on 5 October, and says ChatGPT Ads reached a $1 billion annualised revenue run rate on 31 August. We could not open OpenAI’s own pages for these.
On referrals, Position Digital reports that from about 8 October ChatGPT stopped adding utm_source=chatgpt.com to links, and that OpenAI confirmed the bug and fixed it by 15 October. If your analytics showed a dip in ChatGPT referrals in that period, this may be why, though we have not verified it.
The digest also says ChatGPT referrals from Reddit citations dropped around 80 percent around 14 August, per Petra Labs, and that OpenAI closed its ChatGPT Atlas browser in August, moving its browser proxy function into ChatGPT and Codex. It reports an OpenAI safety check called “Aura” that can block sites from ChatGPT browsing even when robots.txt allows crawling, with no public appeal route. These are second-hand and one-source claims, so verify them before acting.
Microsoft Bing and Clarity
According to Position Digital, Bing now lets publishers separate AI training from AI grounding. It says grounding controls arrive on 15 December, new preferences will be honoured from 1 February 2027, and a NoArchive and NoCache deadline falls on 1 March 2027. It also lists a Bing Topics Portfolio in beta that groups grounding queries into topics and highlights where competitors are cited but you are not. For Microsoft Clarity, it lists a customisable branded-query setting in the AI Visibility dashboard (October), Page Classification in the Bot Activity dashboard (September) and Agentic Analytics that monitors supported AI agents and simulates whether they can complete tasks, starting with commerce and travel (November).
What it means: if you manage a site, these controls and dashboards are worth reading in Microsoft’s documentation, since the dates are in the future and may change.
Perplexity
Perplexity’s own blog index, which we opened, lists posts including “Building shared workspaces for people and agents” (12 October), “Open-sourcing SPACE, the sandbox platform behind Computer” (15 October), “Teammates are agents for complex, ongoing work” (4 November), “Web grounding for the Decisions API” (4 November) and “Improvements to every Agent API effort level” (5 November). We read titles and dates only. Everything PR separately reports, citing third-party trackers and press, that Computer gained automations and inline charts on 5 October and a local-first Mac mode on 20 October for Pro, Max and Enterprise subscribers with at least 36 GB of unified memory. Position Digital noted no Perplexity-specific search items in its August to November digest.
Publishers, licensing and AI training controls
| Item (reported) | Date | Detail as reported |
|---|---|---|
| Shared AI training opt-out token | October | Google, Microsoft, OpenAI and Anthropic will support one robots.txt token; Google plans support in early 2027 |
| Google publisher payment pilot | September | Google is reportedly paying about 100 publishers, with payments varying from under $1,000 to over $1 million a year for one participant |
| Cloudflare monetisation tools | October | Monetization Gateway (closed beta) and Pay Per Use (beta) let sites charge AI agents or get paid when AI uses content |
| Cloudflare and OpenAI pilot | October | Cloudflare signals used to help OpenAI’s crawler find new pages for ChatGPT search |
| DeepSeek Search API | October | Opened to developers at $0.75 per 1,000 successful requests |
All rows come from Position Digital’s digest, and we did not open the underlying announcements. The payment figures in particular are reported, not confirmed. If you are weighing legal or contractual choices about AI access to your content, get qualified advice. This article is general information only.
Timeline at a glance
| Date (2026) | Who | Item (as reported) |
|---|---|---|
| August | Search Console generative AI performance reports roll out worldwide | |
| August | OpenAI | ChatGPT Atlas browser closed |
| September | September spam update begins | |
| 8 to 15 October | OpenAI | ChatGPT referral tag bug, then fix |
| 13 October | Gemini 4 Argon for AI Pro and Ultra subscribers | |
| 15 October | “More ways to connect with the web from Search” | |
| October | Multiple | Shared robots.txt token for AI training opt-out announced |
| October | Alphabet | Q3 results: search revenue up 12 percent year over year |
| Early November | Search Console click spikes reported; Reddit CEO comments on referral decline | |
| 15 December | Microsoft | Bing grounding controls (upcoming) |
What it means for tool buyers and site owners
- Do not rely on a single analytics source. A tagging bug, bot traffic and changing reports all appeared in these digests. Compare Search Console, analytics and server logs before you draw conclusions.
- Check your crawler rules. Training opt-out and grounding controls are becoming separate choices. Read each vendor’s documentation and decide what you want for each.
- Expect ads in AI answers. Both ChatGPT and Google AI Mode are reported to be expanding ads, which affects how you read answers and how you buy visibility. See how to read an AI tool’s pricing page for related cost questions.
- Verify what an answer engine tells you. Sources shown in an AI answer can be incomplete or wrong. Use our checking guide and read how hallucinations arise.
- Choose tools on your own tasks. Use our fair comparison method before moving your research habits to a new tool.
What to watch next
These questions come from the sources and are not forecasts.
- Whether Google’s announced support for the shared training opt-out token arrives in early 2027 as planned.
- Whether Search Console adds click and query data to its AI performance report.
- Whether Microsoft’s 15 December grounding controls ship as described.
- Whether the reported OpenAI business-visibility tool, which a digest says is being built with no launch date, is ever released.
- Whether first-party sources confirm the second-hand items above.
Frequently asked questions
Did Google change AI Mode in October 2026?
A digest reports that Google published a post on 15 October describing inline links, a test section for official websites, expanded news carousels and creator profiles in AI Mode. Confirm on Google’s blog.
Can I opt out of AI training with robots.txt?
A digest reports a shared robots.txt token supported by Google, Microsoft, OpenAI and Anthropic, with Google planning support in early 2027. We could not confirm the exact details on vendor pages.
Are there ads in ChatGPT?
Digests and a release-notes timeline report ChatGPT Ads, including availability in 89 countries in October. Check OpenAI for current rules.
Why did my ChatGPT referral traffic dip in October?
One digest reports a bug from about 8 October that dropped the referral tag, fixed by 15 October. We have not verified this independently.
Does AI Mode hurt site traffic?
The sources we read include a report that Reddit’s CEO said Google referrals fell meaningfully in the quarter. That is one company’s comment. We found no independent measurement.
Which answer engine should I use?
We cannot say, as we have not tested them. Compare them on your own questions and check the sources they cite.
Conclusion
The AI search picture this quarter is one of new controls, new ads and changing measurement, with much of the detail still coming from industry trackers rather than vendors. If you publish content, verify any reported change on the vendor’s own page. If you use these tools for research, choose carefully, check answers, and protect company data.
Sources
- Position Digital, AI search news, August to November 2026, accessed 11 November 2026
- Everything PR, AI answer engine update digest (October 2026), accessed 11 November 2026
- Perplexity blog index, accessed 11 November 2026
- ScriptByAI, OpenAI and ChatGPT timeline, accessed 11 November 2026
