OpenAI posted a senior engineering role on its Monetization team focused on building new ad formats for ChatGPT, with a salary range of $230,000 to $385,000, according to Let’s Data Science, which reported on the listing on July 1. The role explicitly requires building image, video, native, and conversational ad rendering with privacy-preserving, policy-aware design.
The hire comes five months after OpenAI expanded ChatGPT ads from the US to the UK, Brazil, and Japan in May and signals a move from single-format ad units to a broader in-chat advertising suite.
From Engineers to Ad Formats
OpenAI’s ad infrastructure buildout has followed a deliberate sequence. In January, Digiday reported that OpenAI’s seven initial ad-focused roles were almost entirely engineering positions: monetization infrastructure, ad delivery, and internal tooling. No ad sales chiefs, no account managers. The company was building plumbing before hiring salespeople.
The new Ad Formats role represents the next layer: designing how ads actually appear inside conversational interfaces. Image and video ads in a chat window are structurally different from banner ads on a webpage. Conversational ad formats, where advertising integrates with ChatGPT’s response flow, raise harder design problems around user trust and response integrity.
The Agent Infrastructure Problem
The job posting’s emphasis on privacy-preserving rendering points to a problem that extends beyond ChatGPT’s consumer interface. As AI agents become primary interfaces for tasks like research, shopping, and scheduling, advertising infrastructure must work without poisoning agent reasoning or leaking user data to ad networks.
An agent executing a multi-step purchasing workflow, for example, cannot have its recommendation logic distorted by ad placement incentives. An ad rendering system that intercepts agent context to serve targeted ads would compromise the trust model that makes agent delegation work. The privacy-preserving requirement in the job posting suggests OpenAI is designing for this constraint from the start.
“The economics of LLMs just don’t work without some monetization layer, and most users aren’t going to pay,” Maor Sadra, CEO of ad measurement business INCRMNTAL, told Digiday in January. “So what do you do? You offer premium features for ‘free’ and subsidize them with highly targeted ads.”
Early Friction
The path has not been smooth. By May, Digiday’s analysis showed ChatGPT’s average daily time spent per user had dropped 18.3% between March and May (from 25.2 to 20.6 minutes), even as advertiser sentiment improved. Public reaction followed what Digiday described as a “backlash-to-acceptance curve,” starting at -17.9% net sentiment at launch and reaching 49.3% by mid-April.
The engagement dip matters for an ad business. Fewer minutes per session means fewer ad impressions per user, which means OpenAI needs either more users or higher-value ad placements to hit revenue targets. The shift toward richer formats (image, video, native) is a standard response: higher CPMs per impression compensate for declining session time.
Revenue Pressure
OpenAI’s ad ambitions exist within a broader revenue gap. The company closed a $122 billion funding round on March 31 at an $852 billion valuation. Subscription revenue from ChatGPT Plus and Enterprise does not close the gap between inference costs and investor expectations. Advertising is the most proven path to consumer-scale monetization in tech, and Fidji Simo, OpenAI’s CEO of applications and a former architect of Facebook’s News Feed ad strategy, has the playbook.
Whether that playbook translates to conversational AI, where user trust and response quality are the core product, remains the open question. The new Ad Formats engineer will be among the first to find out.