OpenAI's Brad Lightcap exits after eight years to start something new

OpenAI's longest-serving business executive is out the door, and the shakeup at the world's most valuable AI lab just got its biggest name yet.

Brad Lightcap, who spent eight years at OpenAI — most recently leading its special projects division after years as chief operating officer — announced Tuesday he is leaving the company "to start something new." "Team, it is bittersweet to share that I'll be moving on from OpenAI to start something new," Lightcap wrote on X, without naming his next venture. The Information reports he had been running the special-projects group since April, when OpenAI handed most of his COO duties to chief revenue officer Denise Dresser — so the operational handoff was already in motion before this announcement. CNBC calls the exit "the latest in a series of recent leadership shakeups" at the lab.

The departure lands in a year of unusual turnover. It follows the quiet exit of Chloé Bakalar, OpenAI's only ethicist, after less than a year on the job — we covered that this morning — OpenAI's only ethicist exits after less than a year. And it comes days after the company closed a $7 billion tender offer at an $852 billion valuation, with employees and investors alike watching which of the old guard stays for the eventual public listing.

What to make of it: Lightcap was the rare operator who bridged OpenAI's research culture and its commercial machine — the executive who drove the enterprise push and global expansion that turned a research lab into a real business, per CNBC's coverage of his tenure. His exit matters less for day-to-day operations, which Dresser already runs, than for what it signals: the people who built OpenAI's first act are moving on just as the company enters its second, as a for-profit giant heading toward a public market debut. And "start something new" is the exact phrasing of a growing alumni wave — former OpenAI and Anthropic staff have founded dozens of startups this year, and Lightcap's dealmaking record makes his next act one of the most-watched in the industry.

What to watch: what Lightcap builds next — and whether more of the founding-era leadership follows him out before the IPO.

OpenAI keeps losing the executives who built its business — who do you think fills the gap? Tell us in the comments.

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Microsoft finished pre-training its biggest in-house model yet — then used it to make a point about the company it bankrolls. Microsoft AI completed pre-training of MAI-Base-2, a mixture-of-experts foundation model with 80 billion active and 2.8 trillion total parameters, and says the recipe is 4.2–12.3x more pre-training FLOPs-efficient than DeepSeek-V4-Pro-Base. The run itself is the number worth keeping: 16,000 Nvidia GB200 GPUs for 38 days across August and September, at 21% average model

Amazon's new Bee devices listen all day — for $11.99 a month

Amazon turned its always-listening Bee line into a two-device family on Thursday, and NVIDIA opened a 2.8-trillion-token Indic corpus recipe. The connecting thread: both are bets that the next AI money is in memory — yours, or your model's. Amazon put two new always-listening devices on sale, and both now lead to a subscription. Bee Frontier, the next-generation wristband that also clips to clothing, ships now at $74.99 in Onyx, Lavender and Sage — up from the $50 predecessor — with seven-day

Perplexity open-sources its agent sandbox after 118 failed escapes

Perplexity handed its agent containment layer to the open-source world, and Washington's AI task force stopped reading briefings and started working. Perplexity open-sourced SPACE on Thursday — the sandbox it built to keep its own AI agents from getting out — and says nine frontier models failed to break out of it across 118 attempts. Each agent runs inside a small virtual machine with its own operating system, with credentials and network access managed outside the sandbox so an agent can re

Read more

Brain organoids classified Braille at 83% — in a trio, not alone

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Amazon's new Bee devices listen all day — for $11.99 a month

Perplexity open-sources its agent sandbox after 118 failed escapes