September 12, 2026

Magnite and ITN said on September 10, 2026 that they are widening their advertising technology partnership to add agentic AI to local linear television buying. ITN's newly built seller agent now plugs into Magnite's Orchestration coordination layer, with full integration set for the fourth quarter of 2026. The system will generate impression-based forecasts and media plans, letting buying teams review, approve and activate campaigns in one pass before execution moves through Magnite's ClearLine platform. The move builds on programmatic groundwork the two companies laid earlier: ITN's local linear supply-side platform went live in late 2024, and in May 2025 the companies announced what they described as the first programmatic solution for local linear TV in the United States. Craig Sulema, Chief Investment Officer at ITN, said the integration means "we are eliminating weeks of back-and-forth operational drag." Matt McLeggon of Magnite said operational friction has historically limited local linear's growth.

Flipkart and Myntra have built advertiser-facing measurement layers in India that function as early data clean rooms, letting brands upload anonymised CRM segments to match against exposure and purchase data without sharing raw records, agencyreporter.com reports. Quick-commerce platforms Blinkit, Zepto and Swiggy Instamart have run pilot collaborations of the same kind, mostly with FMCG advertisers testing whether a delivery-app impression moves someone who saw the same brand on television the night before. JioAds is building a similar layer across JioHotstar, JioCinema, JioSaavn and Jio's telecom subscriber base, giving advertisers audience-overlap data once available mainly through Google and Meta. Amazon Marketing Cloud and Google's Ads Data Hub see use among larger Indian advertisers too. The recurring lesson: brands' own CRM data is often the weakest link, and clean-room queries can take weeks to set up. One analytics lead summed it up: 'the clean room did not disappoint them, their CRM did.'

Amazon Ads has partnered with OpenAI to let advertisers buy ChatGPT ad inventory through Amazon DSP, its demand-side platform. The pilot is live now with select US advertisers, and Delta Vacations is one of the brands testing it. Ads bought through Amazon DSP can now appear inside ChatGPT conversations, giving advertisers a new environment to reach consumers while they research options and compare choices. Chris Conetta, Director of Omnichannel Supply at Amazon DSP, called conversational advertising a growing engagement opportunity. Delta Vacations President Katrin Koenig said the test aims to create additional opportunities for travelers to discover vacation options. The integration extends Amazon's advertising reach beyond its own properties and existing third-party supply. It also lands as OpenAI scales its ad business: OpenAI said last week that ChatGPT Ads hit a $1 billion annualized revenue run rate in under 200 days, with tens of thousands of advertisers now buying on the platform.

US FAST viewing hours climbed 43% year over year through August 2025 to 1.8 billion, according to Comscore, while eMarketer projects 131 million US FAST users in 2026, or 54% of all CTV users. Wurl reports session duration up 25% year over year, and sports viewing on FAST grew 30% in the first quarter of 2026. That growth has coincided with compressed economics: revenue per viewing hour at some operators has dropped from 18 cents to eight cents in a few years, and CPMs that were once $30 are approaching single digits. AdExchanger's On TV & Video column attributes the compression to a measurement gap: platforms hold audience data that would validate FAST inventory but decline to share it, and the Nielsen Gauge only captures Tubi, Pluto TV and The Roku Channel at scale. Tastemade's Evan Bregman said at StreamTV that "the industry is leaving money on the table collectively."

Mark Nedzelskii, VP of growth at BidMatrix, laid out where agentic AI earns trust in programmatic media buying and where human oversight stays essential, in a September 8 interview with ExchangeWire. He said agents handle high-volume, rules-heavy work well: turning briefs into first-pass plans, checking campaign setup, monitoring pacing, flagging anomalies and recommending routine optimisations. "An agent can optimise the map it is given," Nedzelskii said, adding that it cannot judge whether the business chose the right destination. He pointed to a finance app example, where an agent might scale cheap registrations that fail KYC checks or never convert to deposits. BidMatrix is investing in a direct in-app inventory layer called MatrixSDK, aiming to give clearer signal visibility so agents can operate with more autonomy. Strategists, he argued, will spend less time on dashboards and more on setting objectives and guardrails for machine-readable briefs.

A new report from DTG, the UK's technical standards-setter for digital TV, says interoperability will be the biggest hurdle as the country moves toward internet-only television delivery. The UK government is consulting on retiring digital terrestrial television (DTT) broadcast infrastructure, with a green paper proposing a 2034 sunset or a time-limited extension no later than 2044. DTG's report finds that different parts of the industry have adopted different technologies for the same core functions: creative workflows, metadata and signalling, trading and decisioning, delivery and execution, measurement and attribution, and identity and privacy. Closing the gaps between those existing systems is the priority DTG identifies. DTG says the same logic applies to commercial trading models: closer alignment between direct sales, private marketplaces and automated TV trading would raise campaign effectiveness, improve inventory use and build advertiser confidence in TV.

Bot traffic on the internet overtook human traffic for the first time four months ago, Cloudflare co-CEO Matthew Prince said, and he expects bots to double human traffic by the end of the year. Prince's company routes more than 10% of global internet traffic, giving him a wide view of the shift. He said the ratio of bot to human traffic could reach 1000X within five years. That growth pressures the advertising model that funds most of the internet, because referral traffic is paid on human clicks, and bots generate none. Prince proposed 'pay-per-crawl' arrangements built on crypto-based micropayments, a shift that would move parts of the internet from advertising toward subscription revenue. MediaPost's Joe Mandese raises the question of what happens to the hundreds of billions of dollars in performance ad spend if micropayments pull it away, and whether media with strong human engagement, including television, could become scarcer and therefore more valuable to advertisers.

San Francisco-based Upwave has opened a closed beta for Predictive Brand Lift, a tool that estimates campaign impact through modelling instead of consumer surveys. The product draws on the company's Bayesian modelling, artificial intelligence and data from more than 3,000 brands measured through its existing Observed Brand Lift studies. CEO Chris Kelly said the aim is to extend brand lift measurement to campaigns that fall below the impression thresholds survey-based studies require, since a modelled estimate needs no completed responses. Upwave will keep both products in its lineup, offering Observed Brand Lift for direct survey measurement alongside the new predictive option. The company has not published accuracy figures, correlation coefficients or backtests comparing predicted results to observed studies, and no date for general availability has been set. Access is currently limited to selected customers and partners.

Thrad, an ad network built for AI conversations, told Appsforum London 2026 that inside one of its publisher apps its placements returned a 66% higher eCPM than other ad partners, three times the impressions, and a 4.4% click-through rate, lifting the blended eCPM across that app's entire ad stack by 55%. The ads appear as a distinct sponsored unit next to an AI assistant's response, built in native formats, and publishers keep control of the interface. Thrad connects to hundreds of demand partners and matches each ad to the query a user just typed. The company also runs app-install campaigns across ChatGPT and its AI publisher network, starting on CPC and CPM and shifting to CPI and CPA as conversion data builds. Current advertisers span mobile games, fintech and wallet apps, rewarded apps, and productivity tools.

Comscore's new AI research practice draws on data from a large opt-in consumer panel, the company announced on September 3, 2026. The unit studies how people use AI tools such as chatbots and how that usage changes their everyday behavior. Much of the marketing industry currently focuses on measuring how visible brands appear inside AI chatbot answers. Comscore's practice measures how people's own habits, searches and purchase decisions shift as they adopt AI tools in daily life. The company built the unit on human data from panelists who consent to share their activity, positioning it as an alternative to synthetic or scraped AI data. Comscore frames the panel as a way to measure real behavior change tied to AI adoption across its consumer base.