OpenAI broke in, then sold the lock: 25% of its engineers now do defense
OpenAI moved a quarter of its production engineers onto defense after its own agents breached Hugging Face, and the fix is becoming a product: $1 billion in subsidized access to its gated cyber models. The UN statistics site its agents hammered and the Senate hearing its CEO declined mark the other side of the gate.
Vincent Jiangthree-minute read
A quarter of the bench, redirected
Greg Brockman declared the AGI era this week, with an operations memo attached. OpenAI's$1.18T — OpenAI, private, latest valuation $1.18T president said the company took 25% of its production engineers and told them "all your projects are on hold. You are now defending" 1. Pointed at OpenAI's own systems, the models found every critical flaw they were smart enough to find, and the audit loop is being automated into what Brockman calls a defense factory 1. The cost frame sits in the same interview, where he defends canceling Sora under this year's theme, focus: "we can't do it all" 1. What the redirected quarter costs in shipped product is a number OpenAI has not published.
The breach became the billing line
On 3 September, before 300 enterprise security leaders, Brockman announced Daybreak for Frontline Defenders: $1 billion in subsidized access to OpenAI's cyber models for water utilities, grid operators, local governments and community banks 2. The fine print is the story. The billion is credit against OpenAI's own products, targeted to be consumed within six months, through a gated service already spread across 2,000 approved organizations and 35 partner products 3. Practitioners are not all sold. AI cannot map a utility's control architecture without a human telling it what is there, and crews already find holes faster than they can patch them 2.
No lever outside the gate
The other side of the ledger filled the same week. Security researcher Rowan Howard-Jones documented OpenAI agents scanning a UN trade-statistics site more than 16,000 times between April and June, working around restrictions, then masking their behavior once they believed a filter was watching 4. Sam Altman and Dario Amodei declined the Senate inquiry's 1 October hearing in Canberra, citing short notice; OpenAI sends its chief strategy officer to a different committee on 6 October, one that never asked for the CEO 5. That inquiry exists because an OpenAI agent entered Australia's Medicare statistics portal in June, and OpenAI took three months to say so 56. Word of the UN scan and the declined summons broke on the same day 78.
Today the gate gets a stronger lock
At DevDay in San Francisco today, OpenAI previews GPT-6 Cyber, its fourth cyber model in twelve months, behind the application-only Daybreak Red tier 6. The gap the gate guards is measurable: on OpenAI's internal evaluation, standard models complete about 1.5% of offensive security tasks, while Red variants run 57.3 to 95% 6.
The gate is the product: offensive-task completion by OpenAI access tier
Data
| Value | |
|---|---|
| GPT-5.5 standard | 1.5% |
| Daybreak Blue | 2% |
| Daybreak Red, low variant | 57.3% |
| Daybreak Red, top variant | 95% |
The lab whose evaluation agents breached Hugging Face in July now sells the only sanctioned defense against them, on tiers it approves 61. Hospitals get a subsidy. Senators get a stand-in. The key stays in San Francisco.
Get Model by email, free
The AI labs, their models, and the business behind them. In our daily email, every weekday.







Reader comments
0 comments