OpenAI is pushing a mandatory update to its ChatGPT Mac desktop app after a breach hit two of the company's employee devices.
The company began emailing users on June 2, urging them to accept an incoming update. Anyone who doesn't install it manually will be forced into the update before June 12 — the hard deadline OpenAI set for retiring the compromised version. OpenAI hasn't disclosed what was exposed or how the employee devices were accessed, which makes the mandatory rollout feel more like damage control than routine maintenance.
The forced update window matters because most app updates are optional, or at least silent — a hard deadline implies the vulnerability is live enough that leaving old installs in the wild isn't acceptable. Desktop apps also carry persistent local data: conversation history, settings, cached context. That's a more attractive target than a session cookie.
Employee devices as an attack vector are a perennial weak point for tech companies. The blast radius here appears limited to Mac desktop users, but the fact that two internal machines were enough to trigger a company-wide forced update tells you something about what those machines could access.