BRIEFING

Channel Weather, Issue 2026-W35: Which platform changed the rules on you this month?

Published by Marketingz, Montreal.

A weekly read on the platforms you rent your demand from. We treat algorithm and policy changes as weather: what shifted, who gets wet, and what a small operator should do differently on Monday. Every claim below links to a dated public source. Where the evidence is thin, we say so.

Which platform changed the rules this month?

Google released the August 2026 spam update on 18 August and marked it complete on 21 August, a run it describes on its own status page as "2 days, 16 hours," applied "globally and to all languages" (Google Search Status Dashboard, incident detail, retrieved 28 August 2026; dashboard history, retrieved 28 August 2026).

Meta and TikTok did not ship one dramatic change this month. They kept tightening screws set earlier in the year, and those still land on ad accounts now. The large mailbox providers did nothing we can find a dated source for, and that quiet is its own kind of forecast.

What did Google's August 2026 spam update change?

It enforced the spam policies already published rather than introducing new ones. Google named no new rules and published no blog post, so the update re-applied standing policy on scaled content abuse, site reputation abuse, expired domain abuse and cloaking.

This is the third confirmed spam update of 2026, after March and June. Google published no blog post and named no new spam policies to go with it. So what the update enforces is the policies already on the books: scaled content abuse, site reputation abuse, expired domain abuse, cloaking, and the rest (Search Engine Journal, 18 August 2026; Google spam policies documentation, retrieved 28 August 2026). For scale, Search Engine Journal notes the March update finished in about 19 hours and 30 minutes and the June update in about 2 days and 1 hour, so this one ran slower than both.

Who gets wet. Sites leaning on high-volume, low-effort content. "Parasite" setups where a trusted domain hosts third-party content on rented subfolders. Reactivated expired domains trading on their old authority. Those are the named targets of the standing policies this update enforces (Google spam policies documentation).

What we cannot tell you. Google never publishes which spam categories a given update weights, so anyone claiming to know exactly what this one hit is guessing. Third-party rank trackers are not your traffic either. Treat a tracker chart as rumor and your Search Console data as evidence.

On Monday, open Search Console. Compare the seven days after 21 August with the seven days before, split by query and by page rather than sitewide. If you are down, look for the patterns the spam policies describe on your own pages before you decide it is a false positive. Recovery from a spam demotion is tied to Google's next assessment cycle, not to an overnight fix, so do the cleanup now and expect a wait.

Does Meta now label AI-generated ads automatically?

Yes. Meta applies an "AI info" label automatically when you use its own generative tools, and when it detects third-party AI through C2PA provenance metadata. Ads showing a photorealistic AI-generated person carry the label beside "Sponsored".

The change still working through ad accounts was announced by Meta in June 2026 and covered in trade press on 9 July 2026. Meta's "About this ad" panel now carries an "AI info" label. It is applied automatically when you use Meta's own generative AI creative tools, and when Meta detects third-party AI tools through C2PA provenance metadata. Ads that contain a photorealistic AI-generated person get the label placed next to "Sponsored" rather than buried in a menu (Social Media Today, 9 July 2026). The underlying transparency policy for AI in ads dates to February 2025 (Meta newsroom, February 2025). So this is a mid-year widening of detection and placement, not a fresh August rule.

Who gets wet. Anyone running AI-generated product shots, AI backgrounds, synthetic voiceover, or AI-generated video. Performance advertisers who churn through large volumes of test creative are the most exposed, because they have more assets that could trip detection.

What we cannot tell you. Whether Meta's third-party detection is accurate at scale, or how often it mislabels. It is early, and Meta has said its approach is still changing. The label is not a penalty in itself. The real question is what a visible "AI info" tag does to response for a given audience, and you only learn that by measuring it.

On Monday, inventory your live creative and mark every asset that used generative AI for anything a viewer sees or hears. Where the option exists, set your own disclosure rather than let detection decide for you. Add a labelled-versus-not split to your creative reporting for the next few weeks so you are measuring the effect instead of guessing at it.

Which ad categories did TikTok add rules for?

In August 2026: alcohol, lotteries, NGO documentation, precious metals trading, and contracts for difference. July 2026 added financial services, investment advice, prediction markets, weight loss and muscle gain, and crypto mining devices.

TikTok's public Ad Policy Change Log shows August 2026 entries for Alcohol, Lotteries, an NGO "MDA" requirement update, Precious Metals Trading, and CFDs, or contracts for difference. July 2026 added Financial Services, Investment Consultation or Advice, Prediction Markets and Event Contracts, Weight Loss and Muscle Gain, and Crypto Mining Devices, among others (TikTok Ad Policy Change Log 2026, retrieved 28 August 2026). The log dates entries by month, not by day, so we cannot give a precise effective date for each one.

This is the backdrop, not this month: since 22 January 2026 the US app has run under a new joint venture led by Oracle, Silver Lake, and MGX, with ByteDance holding a minority stake, and the venture has been retraining the recommendation model on US user data (Al Jazeera, 23 January 2026). Whether that has moved organic reach or ad delivery is not something public data lets us show yet.

Who gets wet. Advertisers in finance and trading, supplements and weight loss, alcohol, and anything lottery-shaped or contest-shaped. Expect more documentation requests and more pre-approvals before a flight can run.

On Monday, if you advertise in any category above, open the specific policy page linked from the change log and get your paperwork ready before your next flight, not after a rejection. If you use AI creators or avatars, keep them clearly fictional or clearly consented. Trade coverage says TikTok treats undisclosed synthetic endorsements as prohibited while still allowing purpose-built AI avatars, but that distinction sits in secondary reporting more than in a dated primary entry, so treat it as direction rather than settled fact.

Did the email sending rules change this month?

No. We found no dated change from Gmail, Yahoo or Microsoft consumer mail in the last 30 days. The rules in force are the 2024 bulk-sender requirements: SPF, DKIM and DMARC, one-click unsubscribe, and spam complaints under 0.3 percent.

Nothing shifted that we can source. We checked for a dated change in the last 30 days from Gmail, Yahoo, and Microsoft consumer mail, and did not find one.

The rules in force are the ones set earlier. Google and Yahoo's bulk-sender requirements, which they began enforcing in 2024, ask senders to authenticate with SPF, DKIM, and DMARC, to offer one-click unsubscribe, and to keep spam complaints below 0.3 percent (Google Email sender guidelines, retrieved 28 August 2026). Microsoft matched this for senders above 5,000 messages a day to outlook.com, hotmail.com, and live.com, and moved to outright rejection with a "550 5.7.515" error on 5 May 2025 rather than routing that mail to junk (dmarcian summary of Microsoft's announcement).

We have seen unverified chatter that Gmail tightened enforcement thresholds in late 2025. We could not confirm it against a primary source. Flagging it as a maybe.

Who gets wet. Small senders who assume the 5,000-a-day threshold means "not me." The threshold counts mail to consumer inboxes, and one busy campaign plus routine transactional mail can cross it in a day. Whether your authentication matches your sending domain, and your complaint rate, are judged whether or not you are formally classed as a bulk sender.

On Monday, confirm that SPF, DKIM, and DMARC all pass for the exact from-domain you send campaigns from, and that DMARC matches that domain. A DMARC report reader will tell you within a day. Confirm one-click unsubscribe is live in your marketing sends. Look at your complaint rate, and if it is anywhere near 0.3 percent, cut your least engaged segment before it costs you the inbox.

What could we not pin down?

What should you watch next?

Google may follow this with a core update. It has confirmed spam updates in March, June, and August 2026, and a broad core update would show on the same Search Status Dashboard. Watch that page, not the trackers.

On Meta, watch whether the "AI info" label spreads from the transparency panel onto the face of more ad formats, and whether Meta ever publishes rejection or appeal figures for AI creative.

On TikTok, watch the individual category policy pages linked from the change log fill in with real detail, and watch for any first credible public read on delivery under the new joint venture.

On email, watch for a dated update to Google's or Yahoo's sender guidelines, and for any sign Microsoft is extending its rejection regime below the 5,000-a-day line.

Channel Weather is one of eight segments in this letter and returns in eight weeks. Next week: The Cost of Delay, on how long money is taking to move and what the wait is worth.

Questions, or a correction: marketingz@campaigncrush.org or (305) 765-4411.