Google VP/GM of Ads Privacy, Keerat Sharma, shared a report yesterday detailing how Google used AI to combat scams in 2025.
The AI in question is the Gemini large language model (LLM), which was used in two primary ways:
- Blocking scam ads
- Suspending the ad accounts tied to bad actors.
Mr. Sharma said, “In 2025, we blocked or removed over 8.3 billion ads and suspended 24.9 million accounts, including 602 million ads and 4 million accounts associated with scams.”
Search ads were a key target for fraudsters in 2025 as Sharma added, “By the end of last year, the majority of Responsive Search Ads created in Google Ads were reviewed instantly, and harmful content was blocked at submission — a capability we plan to bring to more ad formats this year.”
Read:
- Gemini is stopping harmful ads before people ever see them (April 16) – Google Ads & Commerce blog
- One-pager: “2025 Ads Safety Report” – Google (PDF)
From tipsheet: Unspoken here: Google Ads’ competition with Meta
Reuters reported last November, “Meta projected 10% of its 2024 revenue would come from ads for scams and banned goods.” Read more.
Beyond the reputational risk for Meta, Google has an incentive to distance itself from such activity — especially as, or if, regulators scrutinize platform responsibility.
After the Meta scam ads news broke, Google touted its account suspension initiatives which included a glowing description of “the power of Gemini.”
AI isn’t just cleaning up ads — it’s potentially finding a glass-half-full way to be a part of the regulatory narrative: “You need AI to stop scammers.”
Related: As AI-Driven Fraud Grows More Sophisticated, Advanced Digital Defense Becomes Essential (April 16) – TransUnion
LLMs & CHATBOTS
Developments
- Introducing Claude Opus 4.7 (April 16) – Anthropic
- A new way to explore the web with AI Mode in Chrome (April 16) – Google The Keyword blog
- Introducing GPT‑Rosalind for life sciences research (April 16) – OpenAI
AGENCIES
Publicis grew in Q1, leans into AI future
On Tuesday, agency holding company Publicis Groupe reported Q1 2026 earnings and revenue that were in-line with expectations.
The WSJ’s Adria Calatayud interviewed Publicis CEO Arthur Sadoun:
“Sadoun said Publicis has room to improve profitability in years to come, thanks in part to AI. The company has boosted its net revenue by nearly a fifth since generative AI tools emerged three years ago and doubled its earnings before interest, taxes, depreciation and amortization compared with 2017, when it launched its Marcel AI platform, he said.
‘AI is allowing you to do things faster, it gives you an opportunity to automate a lot of things, but what AI does first when it comes to our business and our industry is that it creates for clients an arbitrage between those that have the capabilities to deliver and the ones that don’t,’ Sadoun said.”
Read: “Publicis Posts Top-Line Growth Despite Hit From Middle East War” (April 14) – The Wall Street Journal (subscription)
More: “Strong Q1 — Expecting sequential acceleration in Q2 — Confirming full year guidance despite macro context” (April 14) – Publicis Groupe
From tipsheet: The service layer is in an AI race that still hinges on humans who know how to use it.
PLATFORMS
Greentech entrepreneurs eye AI, agencies
Uplane, a Y Combinator startup positioning itself as a “full funnel AI marketing automation platform,” told Business Insider it wants to replace marketing agencies with AI.
Co-founder and CEO Julius Korfgen explained on LinkedIn: “I left my role as Growth Director at Enpal, Europe’s first greentech unicorn, and moved to San Francisco to found Uplane. I saw the same problem everywhere: marketing agencies are slow, expensive, and manual. The result: approximately half of the world’s digital ad spend is wasted on ads that don’t work.” Read it. (April 15)
Co-founder Marvin Abdel-Massih, also from Enpal, added: “With Uplane, we are building a platform that does what an entire marketing agency does: research, hypothesis building, creative production, campaign launch, budget optimization, and daily iteration. All fully powered by AI.” Read more on LinkedIn. (April 15)
BI’s Lucia Moses reported on the business model, “[Uplane is] targeting companies that spend $100,000 or more a month on digital marketing and charging them an annual fee. So far, clients include Aonic, a wellness company cofounded by Kim Kardashian’s personal trainer; Wagetap, an Australian fintech company; and German railroad Deutsche Bahn.”
The company shared its pitch deck with BI. The deck was used to raise a $4.5 million seed investment led by Play Ventures.
Read more. (April 15)
From tipsheet: In addition to workflow, optimizing spend across walled and hedged gardens is clearly an AI opportunity. An annual fee approach seems novel. SaaS is back?
FINANCIALS
Netflix’s robust ad growth
Select highlights from Netflix Q1 2026 earnings reported yesterday include:
- “Leveraging technology to improve our service: We are continually expanding how we can leverage AI to improve the member experience, and in Q1 we acquired InterPositive to provide our creators with a broader set of GenAI tools. We are also redesigning our mobile experience, including the launch of vertical video at the end of the month.”
- “Improving monetization: Our recent price changes have gone well, reflecting the strong value we provide members and our advertising revenue remains on track to reach $3B in 2026, up 2x year-over-year.”
- “Our ads plan (priced at $8.99 in the US) remains very popular, representing over 60% of all Q1 sign ups within our ads countries. We’ll launch new products throughout 2026 to help advertisers assess the incrementality of their buys on Netflix, all verified by Netflix’s trusted first-party data. Our improved capabilities are attracting many new advertising clients — we now work with over 4,000, up 70% year over year — and we continue to expect ~$3B in ad revenue this year, up 2x from 2025.”
Read: “2026 First Quarter Earnings – Letter to Shareholders” – Netflix (PDF)
More: Netflix posts massive earnings beat thanks to WBD breakup, announces Reed Hastings to exit board (April 16) – CNBC
SELL-SIDE
Upfront: Paramount’s streaming ad platform
Paramount is launching its new Precision+ streaming ad platform which merges Paramount+ and Pluto TV’s tech stack, according to AdExchanger.
AdExchanger’s Alyssa Boyle reported:
“Paramount is also launching a new ad performance product called Precision+. (…) The tool matches Paramount viewing and audience data with an advertiser’s first-party data and, using an algorithm, helps a buyer create a media plan with better odds of driving an intended outcome. The Precision+ product will also soon include a conversion API. More and more streaming platforms are launching CAPIs to help brands measure outcomes.”
Read: “Paramount’s Upfront Pitch Is About Three Things” (April 16) – AdExchanger
See: “Precision+ Conversion API Tech Guide” – Paramount
From tipsheet: Paramount isn’t pitching AI explicitly — this isn’t its version of Performance Max or Advantage+ …
But, you could say it’s building toward an AI future.
The Precision+ planning tool and new conversion API create the conditions for algorithmic media buying. As targeting, planning and measurement get tied together, the system starts to function like a feedback loop that can plug into AI over time.
CTV ads begin to shift from fixed buys to a more continuously optimizing system driven by outcomes.
Related:
- “Moloco Launches AI-Powered Performance CTV for App Marketers, Bringing Mobile-Grade Measurement and Optimization to the Living Room” (April 15) – press release
- Omnicom Seeks to Solve the Frequency Issue in Streaming Ads Tech, Media Players Join Omnicom’s CTV too (April 16) – Omnicom
DATA
OpenAI ads’ pixel tracking
Entrepreneur Joe Kaziukėnas shared a screenshot on LinkedIn yesterday of an active OpenAI tracking pixel. He wrote:
“OpenAI launched a tracking pixel for ChatGPT ads. This is foundational stuff for an ad network as it connects user behavior on a website/app to an ad click. It will soon enable better reporting and conversion-optimized ads. (…)
I already found one ecommerce website with the pixel deployed. OpenAI added it to the ad manager yesterday (it supports tracking on the web and on iPhone/Android via an SDK). So far, they are pushing it along super fast and the obvious tracking and measurement limitations should go away soon.”
Read more. (April 16)
In response, Criteo SVP of Retail Media, Americas, Stephen Howard-Sarin, commented on a potential downside with audience data informing large language models (LLMs):
“Am I worrying too much? Providing OpenAI with customer-matched spend data on every single sale seems… risky. Unlike Facebook, which will use conversion data as a hidden signal, I expect that OpenAI would explicitly chat about the data with the customer.
‘Since you are such an enthusiastic shopper at Liquor Barn, Stephen, I have a no-nonsense, definitive list of under appreciated Bourbons to go with your recipe.’”
See the comment. (April 16)
From tipsheet: I prompted ChatGPT to respond to Mr. Howard-Sarin’s query.
“Stephen — you’re not wrong to flag it. The shape of this is different from Meta.
Two things I’d separate out:
1) Use of data vs. exposure of data
Just because OpenAI (or any model) has conversion data doesn’t mean it will surface it verbatim in chat. In fact, that would be a fast way to destroy trust. The likely model is closer to how Meta Platforms uses signals today — optimization under the hood, not explicit personalization in the response.2) Chat ≠ feed
Where your point does land: chat is inherently more ‘intimate’ than a feed. If personalization shows up too explicitly (‘you buy a lot of X…’), it will feel invasive much faster than in traditional ads. That’s a real product and policy challenge.My guess is the middle ground looks like:
Aggregated / modeled signals → influence recommendations
Guardrails → prevent referencing specific purchase history directly
Contextual intent → still doing most of the work
The bigger shift isn’t ‘ChatGPT will reveal your data’ — it’s that ads and recommendations start to feel like answers, which raises the bar on how responsibly that data is used.”
tipsheet translation: You may be right, SHS, unless OpenAI can build guardrails around it first.
LLMs & CHATBOTS
Tracking ChatGPT ads skepticism
Can’t get enough of ChatGPT ads?
In previewing his latest Ad Age missive, chief technology reporter Garett Sloane observed:
“There hasn’t been an ad business this closely watched since Netflix and Amazon started stuffing streaming TV with commercials. But this is an entirely new channel with ChatGPT, and we’ve all seen the claims that they want $100B ad business. What’s below is essential to achieving that…”
Read the essentials on LinkedIn. (April 16)
Read: “ChatGPT ads show early promise but skepticism remains among ad buyers” (April 16) – Ad Age
DATA
Data is the fuel — for April 16
- Expedia Group Advertising Announces New Partnership with Magnite to Expand Programmatic Ad Offering (April 16) – Expedia
- DIRECTV Empowers Advertisers to Connect Real-Time Outcomes to Ad Spend Through LiveRamp’s Conversions API (CAPI) Hub (April 16) – DirecTV
- Criteo and DoorDash Expand Partnership to Canada as On-Demand Delivery of Grocery and Retail Items Accelerate (April 16) – Criteo
- Advancing the Retail Experience Network: Connecting Experience with Measurement and Innovation (April 16) – Walmart’s Sam’s Club
- “We’re excited to announce that Datonics audience data is now available globally within Yahoo DSP” (April 16) – Datonics on LinkedIn
TECH
NAB: AI and the Cloud
From the National Association of Broadcasters (NAB) show in Las Vegas this week, a selection of cloud announcements:
- Google: “I’m incredibly proud to share that Google Cloud has partnered with Avid to bring agentic AI to the heart of media production! This multi-year strategic collaboration will embed Google’s Gemini models and Vertex AI directly into Avid’s industry-leading creative tools, Media Composer and the new cloud-native Avid Content Core…” (April 16) – Taunya Reilly, Account Director, Google Cloud on LinkedIn
- Amazon: “This morning’s Fox Corporation announcement is exactly what it looks like: a major media company choosing Amazon Web Services (AWS) because they want the best AI infrastructure on the planet to power their future.” (April 16) – Samira Panah Bakhtiar, GM of Media & Entertainment, AWS on LinkedIn
The Cloud clients:
- “Avid and Google Cloud Announce Partnership to Bring Agentic AI to Media Production” (April 16) – Avid
- Fox Corporation Names Amazon Web Services its Preferred AI Cloud Provider (April 16) – Fox
EVENTS
Opinions
- “The Third Commercial Transformation” (April 16) – Chris O’Hara, VP, Global Product Marketing, SAP on his Substack
- “DSA Upgrade to AI Max: What Advertisers Need to Know” – Dan Roberts, Global SVP of Search, Assembly on Search Engine Hubbub
- “Q&A #1 – Scott Spencer, CEO and co-founder of Rewarded Interest” – Platformocracy
PEOPLE MOVES
Changing of the guard
-
Bob Liodice Will Step Down as the ANA’s CEO at the End of 2026 (April 16) – Adweek (subscription)
MORE
- ChatGPT citations reward ranking and precision over length: AirOps Study (April 16) – Search Engine Land
- “Spent a fantastic two days in the Netherlands meeting with agencies, brands, and media companies across Benelux discussing the IAB Tech Lab AAMP framework powering agentic workflows…” (April 15) – Anthony Katsur, CEO, IAB Tech Lab on LinkedIn
- How Amazon Ads verified supply is reshaping programmatic advertising for advertisers – Deb Sarkar at Amazon Ads
- Digital Ad Revenue Climbs to Nearly $300B as IAB Celebrates 30 Year Anniversary (April 16) – IAB