Three separate product notices landed within about thirty-six hours, none of them presented as a policy shift, and together they describe the same thing: the number of places where a media buyer can name a maximum price is shrinking.

Microsoft moved first, and most explicitly. From October 1, 2026, new campaigns using Max Conversions, Max Conversion Value and Max Clicks will no longer accept a Max CPC. The change applies only to non-portfolio campaigns created after that date. Campaigns already running keep the field. Portfolio bid strategies keep it. Target impression share and enhanced CPC keep it too. Navah Hopkins, the Microsoft Ads Liaison, framed the removal around outcomes rather than control, arguing that advertisers using conversion strategies with explicit targets outperform those leaning on what she called legacy controls, and that a Max CPC "can lead to spend pacing irregularities" and override the goal an advertiser has already stated. Microsoft's own language was blunter still: the ability is being removed "in order to simplify bidding." The company added that further updates on Max CPC would follow, which reads less like a footnote than a schedule for the existing campaigns that were exempted this time.

A day earlier, the same principle appeared in a much younger auction. OpenAI preselected a strategy called Maximize results in eligible new ad groups inside ChatGPT Ads Manager Beta, with the help centre page updated around August 12 and the product email circulating in the week beginning August 17. Maximize results sets and adjusts bids on its own, optimising toward clicks or conversions, and it carries an unusually candid caveat in OpenAI's own documentation: the strategy "prioritizes result volume and does not guarantee delivery against a specific CPA, CPC, ROAS, or other cost-efficiency target at this time." The only documented route to a ceiling is the alternative option, Manual: Max bid. No account-level preference exists to change the default, so the choice recurs at every ad group.

What makes the OpenAI version sharper than Microsoft's is what surrounds it. Fixed daily budgets in ChatGPT Ads were converted to seven-day averages in July 2026 with no opt-out, which loosened the constraint at the level of the individual day. View-through conversions arrived on August 18 and 19 but sit outside bidding, billing and CPA calculation entirely, visible in reporting and invisible to the optimiser. Automatic advanced matching became the default on August 17 with a ten-day opt-out window. Each of those is defensible in isolation. Stacked, they describe an account where the daily spend limit is elastic, the bid has no ceiling by default, and part of the measured outcome cannot be optimised against.

Google's contribution was subtler, because Google is further along. On August 21 the company added experimentation and planning features around AI Max, the search campaign layer it has been pushing since 2025. From September 2026, a single A/B test can span multiple Search campaigns at once, letting advertisers measure how budget and ROI target changes move aggregate performance rather than testing one campaign in isolation. AI Max experiments can now run with brand and location parameters left switched on, which removes an awkward earlier condition where testing the automation meant temporarily abandoning the guardrails. Performance Planner gained the ability to forecast the effect of bidding or budget changes and then apply the recommended adjustment directly, without the advertiser executing it. Google also revised the support document "About AI Max experiments," which retains an archived April 2026 version for comparison.

Read alongside the Microsoft notice, the Google update is the mirror image of the same trade. Microsoft is withdrawing a control and offering targets in exchange. Google is expanding the measurement apparatus around a system that already sets prices, so that the effect of the automation can at least be observed at portfolio scale. The instruments improve while the levers shrink.

OpenAI, meanwhile, gave something back on the same day it took something away. A campaign-level platform targeting control appeared with three values: iOS App, Android App, and Web, the last covering both desktop and mobile browsers. Documentation was updated August 20. The setting is optional, functions as an inclusion list rather than a bid modifier, and sits alongside objective, budget, dates, country list and custom audiences. It is, by any competitive standard, an overdue basic. Every major auction has offered device or platform segmentation for more than a decade.

The problem is that the reporting did not move with it. The Insights dashboard still returns two buckets, Mobile and Desktop, with mobile web folded into Mobile. Three buying values therefore map to two reporting values. An advertiser buying only native app inventory sees Mobile rows and nothing that distinguishes iOS from Android. An advertiser buying only Web cannot separate desktop from mobile browser. OpenAI said more detailed platform reporting is planned and attached no date. Splitting surfaces into duplicate campaigns is possible, but each duplicate must clear its own minimum daily spend: $25 in the United States, £15 in the United Kingdom, $25 in Canada, Australia and New Zealand, ¥2,500 in Japan, ₩25,000 in South Korea, R$40 in Brazil, $150 in Mexico, ₹725 in India.

The rationale is easier to see in the measurement plumbing than in the buying interface. Since February, ChatGPT advertisers bought every surface at once while conversion infrastructure split along application and browser lines, pixel and Conversions API for web, AppsFlyer and Adjust for apps, both of which went live in July 2026. Platform targeting lets buying match the measurement stack that already existed. It also happens to be a lever that carries no behavioural personalisation, which matters in the 31 European countries added the week after August 17, where campaigns launched without personalised advertising.

Adoption data remains contested. Similarweb recorded sponsored placements in 26% of ChatGPT responses at a 0.50% click-through rate, with 7,378 distinct advertisers in the week to August 17. Criteo counted more than 2,000 brands via API by June. Adthena, scraping the United Kingdom, recorded zero placements across 169,560 samples in June. Those three findings are not necessarily in conflict, but they do not describe a settled market.

Fifty-three percent, ninety-five percent, and 1,820 combinations

On August 20, 2026, a seven-count class action complaint was filed against Oura in the Northern District of California, San Francisco Division, as case 3:26-cv-08686. The plaintiff, Madison Surber, a California resident represented by Clarkson Law Firm, bought an Oura Ring 4 in gold on May 22, 2025 for $513.68 while in Los Angeles County. Venue rests on the argument that the marketing at issue originated in San Francisco, where Oura Inc. and Oura Health Oy keep offices at 222 Kearny Street.

The counts are familiar: fraud by misrepresentation, unjust enrichment, breach of express and implied warranties, and violations of California's Unfair Competition Law, False Advertising Law and Consumers Legal Remedies Act. The substance is not. At the centre of the complaint sits a single advertised figure, "95% Sleep Staging Accuracy," alongside a claim of "79% agreement with clinical polysomnography," and the argument that neither can be substantiated by the hardware.

The technical section of the filing is where the case becomes interesting to anyone who writes performance copy. Polysomnography, the clinical reference standard, identifies REM sleep through rapid eye movements on EOG, muscle atonia on EMG and EEG desynchronisation, and identifies deep N3 sleep through waves between 0.5 and 2.0 Hz at a minimum amplitude of 75 microvolts. The Oura Ring carries red and infrared LEDs for blood oxygen, green and infrared LEDs for heart rate, heart rate variability and respiration rate, plus a digital temperature sensor and an accelerometer. It has no EEG, no EOG, no EMG and no ECG. What it produces, the complaint argues, is machine learning inference applied to peripheral physiological signals rather than direct measurement of the thing being claimed.

The supporting evidence is a peer-reviewed Nature study published March 19, 2025, covering 45 patients, which found overall stage classification accuracy of 53.18% against polysomnography. In the same comparison, the rings underestimated light sleep by 13.84 minutes and deep sleep by 5.98 minutes, and overestimated REM sleep by 31.56 minutes. Set against an advertised 95%, the gap is not a rounding question.

Oura's own material appears on both sides of the ledger. A blog post dated November 16, 2022 cited the 79% polysomnography agreement figure. Another, dated December 6, 2022, positioned the ring as an alternative to a sleep study. A third, from February 12, 2024, acknowledged that the rings do not measure electrical brain activity or eye movements. Support documentation states the ring "is not the same as a PSG." The complaint's argument is not that the qualification is absent but that it lives in secondary material while the accuracy figure sits on the product page unaccompanied.

The advertising mechanics are what lift this above a routine consumer suit. Counsel reviewed product pages and the Meta Ad Library between August 17 and 19, 2026 and identified roughly 91 active Meta advertisements carrying the disputed promise. The filing then describes Meta's Dynamic Creative system in detail, calculating that automated mixing of images, videos and headlines yields 1,820 possible combinations, and uses that arithmetic to argue that the messaging was uniform in substance even though delivery was individualised. Retargeting is folded into the same theory, with the complaint citing 2018 industry data on a 70% conversion uplift to argue that repeated exposure compounded the deception. The earliest identified promotional post is dated March 3, 2015, giving the campaign a run of nearly eleven years across Meta, TikTok, Instagram, YouTube and X.

A CLRA notice letter went out by certified mail on March 26, 2026. No adequate correction followed, according to the filing, which is the procedural predicate for the damages claim. The plaintiff seeks a nationwide class of United States purchasers plus a California subclass covering four years, with damages exceeding $5 million, restitution, disgorgement, punitive damages, and a corrective advertising injunction requiring what the complaint describes as an affirmative campaign to dispel the public misperception.

Oura Health Oy was founded in Oulu, Finland in 2015, raising $650,000 on Kickstarter and a $2.3 million seed round. By its Series E in 2025 the company was valued at roughly $11 billion, with 5.5 million rings sold and revenue above $1 billion. Jennifer Aniston, Prince Harry and Kim Kardashian are named in the complaint as endorsers.

Two threads from the filing deserve to be noted separately from the merits, which will take years to resolve. The first is evidentiary: the Meta Ad Library is being treated as a discoverable, citable record of what an advertiser said and when, with counsel timestamping their review window down to a three-day span. The second is structural. Dynamic creative optimisation was built to remove the need to decide which combination a given person sees. This complaint reframes that same automation as an aggravating factor, arguing that the advertiser cannot claim a person saw a nuanced version of the message when the system that assembled it had 1,820 versions available and no record presented to the consumer of which one arrived.

Six hundred thousand sources, and a button publishers can host

Google published three personalisation changes on August 20, 2026, announced by Mrinalini Loew, general manager of the Google Search ecosystem, through a blog post and the News from Google account on X, where the thread had gathered 63,100 views at capture. The headline number: Preferred Sources has now accumulated more than 600,000 unique sources selected by readers, an increase of 255,000 since May 27, 2026, when the count stood above 345,000.

The trajectory is worth laying out because the growth is not linear. The feature began as a Search Labs experiment on June 26, 2025, launched into Top Stories for the United States and India on August 12, 2025, and went global on December 10, 2025 with roughly 90,000 unique sources. The May 27, 2026 expansion into AI Overviews and AI Mode coincided with the jump past 345,000. The three months since have added more selections than the entire first eleven months produced.

The most consequential of the three changes is the embeddable button, which lets a publisher place the selection control on its own pages rather than waiting for a reader to find it inside Google's interface. Three implementation paths exist. The standard route is two lines: a script tag loading the library from news.google.com/swg/js/v1/publisher.js placed in the head element, and a div carrying the google-add-preferred-source-btn attribute wherever the button belongs. The library scans for the attribute and renders the badge. Two data attributes control presentation, data-theme accepting light or dark with light as default, and data-lang overriding browser language. An advanced path imports publisher.mjs directly for custom designs, initialises with a theme and language object, and binds an addPreferredSource() method to any click handler, with auto-rendering suppressed by a preferred-sources-control="manual" attribute. For content management systems that cannot execute JavaScript, a deeplink to google.com/preferences/source with the domain as a query parameter achieves the same result. Button assets are downloadable in 16 languages. Clicking adds the source and returns the reader to their position on the page.

Eligibility is drawn at the domain and subdomain level. Subdirectories, including the common /blog/ pattern, are not eligible, which means a large publisher running consolidated verticals under one domain cannot offer readers a granular choice between them.

The second change tunes Discover by free text. Readers tap the three-dot menu on a Discover card in the Google app and describe in ordinary language what they want more or less of; the feed adjusts immediately and retains the instruction. The capability was previously tested as a "Tailor Your Feed" beta with a chat interface, and pairs with a "You asked to see" transparency label first spotted April 24, 2026. The third is topic curation for Google News audio briefings, restricted to the Google News app on Android, drawing on the commercial AI partnership cohort announced December 10, 2025, which includes Der Spiegel, El País, Folha de S. Paulo, Infobae, Kompas, The Guardian, The Times of India, The Washington Examiner and The Washington Post. Both are arriving "in the coming days."

The context that gives these numbers their weight is traffic. Discover accounts for roughly two-thirds of Google's news referrals, per August 2025 research. Page views from Google Search fell 34% in the twelve months to December 2025, with Discover down 15% over the same period. Marfeel data from December 2025 found AI Summaries occupying 51% of Discover positions in test markets. Against that backdrop, Preferred Sources is the only Google surface where a publisher can directly ask for designation and a reader can directly grant it. Everything else is algorithmically mediated.

Google's engagement claim has been stable and unaudited since December 2025: readers who select a preferred source click through to that site "twice as much on average." Liz Reid, vice president of Google Search, repeated the figure on a podcast on June 26, 2026. No baseline is disclosed, and no persistence data accompanies it. Search Console exposes no preferred-source dimension, so a publisher embedding the button can count clicks on the button in its own analytics and cannot see what the designation subsequently produced in Search. The 600,000 figure counts unique sources selected, not readers, not selections, and not sessions.

A claim that blocks the video and pays nobody

YouTube is renaming the Copyright menu in Studio to Claims and merging two enforcement systems that behaved very differently. Content ID copyright restrictions and AI likeness privacy complaints will share one dashboard. Alongside the consolidation, announced August 19, 2026 by a TeamYouTube representative on the Help Center Community forum, four dispute categories arrived for creators contesting a likeness claim: explicit consent, meaning documented permission from the person depicted; parody, satire or public interest; content not altered or made with AI; and claimed content does not appear, for videos that do not contain the specified face or voice at all. Video overview pages will also start displaying the identity of the claim filer and details of the detected likeness. No rollout date was attached to either change.

The asymmetry between the two claim types is the part with commercial consequences. A copyright claim offers four responses: accept it, remove the content, share the revenue, or dispute. A likeness claim offers two, accept or dispute, and carries no revenue-sharing pathway at all. A likeness claim can block a video from being viewed completely, and a blocked video generates zero revenue for anyone, claimant included. There is no strike attached, neither copyright nor Community Guidelines, which means the penalty falls entirely on distribution rather than on account standing.

Two lines in the documentation carry more weight than their length suggests. "Altered content disclosures don't stop claims," and "Privacy settings do not stop claims." The first means the AI disclosure regime and SynthID watermarking, both designed to satisfy transparency obligations, provide no defence against a likeness complaint. The second means private and unlisted videos remain claimable, which reaches directly into the approval workflows brands and agencies run before a public launch. Shorts running between one and three minutes with an active copyright claim are blocked regardless of what the claimant would otherwise have allowed.

The enrolment curve explains the timing. Likeness detection ran as a Partner Program pilot through 2024 and 2025, expanded on March 10, 2026 to government officials, journalists and political candidates, and opened in May 2026 to all creators aged 18 and over. Every expansion enlarges the population able to file. France Televisions adopted YouTube Likeness ID in April 2026 to protect its presenters. Meanwhile the interface has been rebuilt in stages: the Content tab redesigned June 28, 2026 with a consolidated Notices column, an Account Status page added to the mobile app in July, and channel manager copyright notifications on August 14. Article 50 of the EU AI Act became applicable August 2, 2026, with penalties for unlabelled deepfakes reaching 3% of worldwide annual turnover. The Help Center post had drawn 120 likes and 19 replies at capture.

For advertisers, the practical exposure is not the dispute process but the warranty. A creator contract that promises originality does not prevent a third party from filing a likeness claim against the resulting video, and once filed, the only routes out are acceptance or one of the four categories.

The bellwether nobody wants to be

A Digiday examination of Nike's decline, published August 21, 2026 by Sam Bradley, argues against every convenient single-cause story attached to the company: that direct-to-consumer failed, that political controversy did lasting damage, that the advertising budget was cut too far.

The numbers complicate each of those. Nike's stock sat at a twelve-year low at the time of writing, but the running category it once owned is growing, not shrinking. United States running equipment was a $12.9 billion market last year. Performance footwear sales rose 13% in the first half of 2026. Brooks grew revenue 14% over the same period. Nike held 22.9% of the running market last year, a share it built when the alternatives were fewer. On the political theory, the September 2018 Kaepernick endorsement did produce a 73% drop in favourability among Republicans and a 54.8% fall in purchase consideration within that cohort, but both metrics had climbed back by March 2025. And the company spent $41.7 million on World Cup media in the United States, which is not the behaviour of a brand that stopped advertising.

What the practitioners quoted in the piece describe instead is a loss of intelligence rather than a loss of reach. Shane O'Leary, formerly of Droga5 and GroupM, put it as "there are far more factors at play than just the one of the four Ps." The Consumer Direct Offense strategy under former chief executive John Donahoe consumed resources and attention before reversing toward a hybrid retail model, leaving roughly two years of infrastructure disruption ahead of it. Cutting secondary retailer relationships removed distribution, but Leila Fataar of Platform13 argued it removed something less visible: "Those relationships weren't just commercial. They were part of Nike's cultural intelligence system." Justin Cox of MSQ North America pointed to the persona-building partnership with Wieden+Kennedy as a machine that no longer has the same inputs. Michael Pavone of Pavone Group offered the flattest version: "We're not living in the same world we were 20 years ago."

Rob Baiocco of BAM Connection supplied the line that explains why a footwear company's quarterly troubles are appearing in trade coverage aimed at media buyers. "It's being treated as a bit of a bellwether in the industry." Nike is the case study most often cited to prove that sustained brand investment compounds. A twelve-year stock low invites the opposite reading, and O'Leary supplied the uncomfortable corollary directly: "We probably overestimate the actual effect of advertising. It's called a weak force for good reason."

That sentence sits oddly against the week's other filings. Microsoft removed a price control on the grounds that its automated targets deliver better outcomes. OpenAI made volume maximisation the default while conceding that no cost-efficiency guarantee exists yet. Google built better instruments for observing a system that already sets its own prices. Each of those changes assumes the measured effect of advertising is the real one and the practitioner's judgement is the noise. Nike's decade suggests the causal chain is longer and messier than any auction can see, running through retail shelf space, athlete media independence, supply chains and Chinese competition, none of which appear in a conversion report.

Digiday's briefing also collected a few figures that rhyme with the week: 50% of marketers now report comfort with AI agents making autonomous decisions, per the StackAdapt survey published August 19; Semrush found no clear brand dominance in 85% of AI search categories; and 90% of the Baller League's audience watches through YouTube.

Also noted