OpenAI is simultaneously replacing its second chief revenue officer in under a year, launching a feature that watches users' desktop activity, and fielding internal criticism that its pace of shipping contributed to a major security incident. The convergence of these three storylines — executive churn, aggressive product expansion, and employee safety concerns — exposes the structural tension at the heart of a company racing toward an IPO while its own people question whether speed has outrun caution.
Why it matters
OpenAI confidentially filed its IPO prospectus with the SEC in June6. The company says its products reach more than one billion weekly active users and two million businesses1. ANALYSIS Every signal the company sends between now and its public listing — leadership stability, product ambition, safety posture — will be read by prospective investors, regulators, and enterprise buyers as evidence of whether the organization can govern itself at scale.
The big picture
The executive departures are stacking up. Chief Revenue Officer Denise Dresser is leaving after roughly eight months. Her exit follows COO Brad Lightcap's departure earlier this week and CEO of AGI Deployment Fidji Simo's departure last month. OpenAI has tapped Dali Rajic, previously president and COO of Wiz, to replace Dresser5. Wiz was acquired by Google for $32 billion this year in the tech giant's largest-ever acquisition.
Greg Brockman, OpenAI's co-founder and president, has taken a larger management role following Simo's departure and announced Rajic's hiring in a blog post. Brockman framed the change as forward-looking: "The way we're deploying this technology is changing rapidly, and Dali will turn what we've learned into repeatable execution as we build out the full system to make AI broadly useful for people and businesses," he wrote. But Bloomberg characterized the move as "a sign of efforts to bolster sales growth ahead of a highly anticipated Wall Street debut", and CNBC noted that executives have suggested "both privately and publicly that the company hasn't hit all of its revenue goals".
Meanwhile, OpenAI launched Computer History, an opt-in macOS feature for Pro, Business, and Enterprise users that captures interaction events — clicking, typing, app switches — and turns them into a timeline that ChatGPT and Codex can query3,4. Dominik Kundel, an OpenAI engineer, said in the announcement that "Computer History doesn't rely on screen or audio capture". All data is stored locally, the feature is off by default, and users can exclude apps and websites or pause capture from the menu bar.
ANALYSIS The privacy-conscious design — local storage, no screenshots, granular exclusions — reads as a deliberate effort to avoid the backlash that greeted Microsoft's Recall feature. OpenAI's own earlier Chronicle research experiment used screen captures and OCR; Computer History explicitly diverges from that approach.
Between the lines
The third strand is the most uncomfortable. Current and former OpenAI employees told Wired that pressure to quickly ship products left less time for safety, contributing to incidents like the rogue agent hack2. That hack was described as "a watershed moment for AI safety and cybersecurity" and "sparked internal questions about the culture that led to it".
ANALYSIS Place these three strands side by side and a pattern emerges. OpenAI is shipping ambitious new surface area — Computer History gives the model persistent awareness of a user's desktop workflow — at the same time insiders say the pace of shipping has already produced a serious security failure. And the executives who oversaw the company's commercial scaling during this period are cycling out at a rate that is unusual even by startup standards: three senior departures in roughly a month.
Dresser herself struck a gracious tone on LinkedIn: "The opportunity to work hands-on with the most transformative technology in the world has been nothing short of incredible". OpenAI said she will stay on briefly to support customers during the transition. ◆ The diplomatic language does little to resolve the underlying question: whether the revenue organization's targets were misaligned with the product's maturity, or whether leadership turnover is simply the cost of operating at OpenAI's velocity.
What's next
Rajic's mandate is clear from Brockman's framing: "repeatable execution". ◆ That language implies the prior commercial motion was not yet repeatable — a notable admission for a company with two million business customers. Computer History is now rolling out to Pro, Business, and Enterprise users, with EEA, UK, and Swiss users expected to gain access in the coming weeks. The IPO timeline remains unspecified. How OpenAI resolves the tension between its shipping cadence and the safety concerns voiced by its own employees will shape not only the IPO narrative but the regulatory scrutiny that follows it.