A busy week in AI and marketing. Here are the six stories worth your attention, ranked by how directly they affect operators running campaigns, content, or ad spend.
AI search revenue up 321% in 13 months
Research reported by Marketing Week analyzed 1.5 million AI-search-engaged sessions across 175 websites and 20 industries. Revenue attributable to AI search rose 321% over the past 13 months. Transactions attributed to AI search climbed 553%. AI search traffic is growing 14 times faster than organic search, and AI-engaged sessions have increased nearly fivefold.
The practical implication: ChatGPT, Gemini, and similar platforms are no longer edge cases in your analytics. If you’re still lumping them into referral traffic, you’re blind to a channel that’s already converting.
ChatGPT now running visual ads
OpenAI is rolling out a new visual ad format inside ChatGPT that shows products while users generate images. Ads will be clearly labeled and kept separate from generated images. Attribution integrations with AppsFlyer, Adjust, and Triple Whale are launching this week. Testing of the visual format begins in the US later this month.
OpenAI reports ChatGPT now reaches 1.2 billion people each week. Measurement was the main objection holding advertisers back. With attribution tooling now in place, that objection is weaker.

Bain: top 10% of businesses are redesigning whole workflows
Bain’s 2026 Technology Report draws a sharp line between the top 10% of businesses and everyone else. The leaders have moved past deploying individual AI tools and are rebuilding entire workflows around AI. Those companies are seeing 10% to 25% EBITDA growth. The other 90% are largely achieving what Bain describes as small productivity improvements from narrow use cases.
Marketing-specific numbers from the report: well-designed AI transformations can produce 60% to 70% fewer people required per piece of content, 10% to 30% higher revenue through improved conversion, and 2 to 3 times better conversion rates through AI-driven segmentation.
Bain’s message is direct: every month spent running small experiments is potentially a month in which competitors redesign how they work.
TikTok launches agentic tools to push buyers down the funnel
TikTok unveiled a suite of agentic AI tools designed to move consumers further down the sales funnel. Its new Agentic Leads product uses AI-led conversations to capture and qualify prospects through TikTok direct messages and advertisers’ websites. TikTok is also introducing conversational shopping tools and expanding automation across campaign planning and optimization.
Separately, the TikTok Ad Network is expanding to US advertisers, enabling campaigns to reach users across nearly 400,000 third-party apps.
OpenAI launches Dots: persistent AI agents powered by GPT-6 Astra
OpenAI unveiled Dots, a new AI agent product designed to take responsibility for ongoing work rather than waiting to be prompted. Each Dot runs on GPT-6 Astra, has its own cloud computer, connects to more than 4,000 apps, and learns preferences over time. OpenAI is also testing specialist Dots for businesses, effectively assigning an AI agent to a specific function or role.
For marketers and operators, the implied use cases include continuous campaign monitoring, competitor research, prospect qualification, and report preparation, all running without requiring a human to initiate each task.
US executive order renames AI to “Super Intelligence”
An executive order signed on September 29 directs US federal departments and agencies to replace the terms “Artificial Intelligence” and “AI” with “Super Intelligence” and “SI” in official communications, websites, reports, and policy documents.
The complication: superintelligence already has an established meaning in research. It generally describes a hypothetical intelligence that significantly exceeds human capabilities. Researchers have noted that changing the label does not change the technology underneath it.
For context on how naming shapes entire technology categories: in 1955, John McCarthy coined the term Artificial Intelligence specifically to attract funding for a small research gathering at Dartmouth College. The proposal went to the Rockefeller Foundation with the new name and received $7,500. The resulting Dartmouth workshop in 1956 is now regarded as the starting point of modern AI. McCarthy’s naming choice mattered enormously when the category had no established frame of reference. Today, AI is the most recognized technology category on the planet. Renaming it at the federal level is unlikely to move the needle for anyone outside of government documents.
Follow the money: $450 billion in AI-related debt by September
The Bank of England reported that global AI-related debt issuance reached approximately $450 billion by early September 2026. That figure is already more than double the total issued in all of 2025. Morgan Stanley estimates cited by the Bank of England suggest AI-related capital expenditure financed through debt could reach $4.1 trillion between 2026 and 2030.
AI companies have accounted for 47% of all sterling corporate bond issuance so far this year. An estimated $700 billion of data-center investment between 2026 and 2028 could be funded through private credit. The Bank of England now warns that rising debt, opaque financing structures, and circular arrangements could amplify losses if AI’s expected productivity gains fail to materialize. It also cautions that a reassessment of AI’s productivity assumptions could affect not just technology shares but wider financial markets.
