OpenAI يضيف شروط الترخيص على البريد الإلكتروني التي تتطلب إزالة كاملة عند الانتهاء

OpenAI’s Service Terms now include a section for “Licensed Materials” that defines software delivered for installation on a customer’s own machines, and the clause every procurement team will hate: on termination, you must permanently delete every copy.

That language isn’t boilerplate. Contract terms for on-prem delivery tend to ship before the actual product does, which means OpenAI is laying the legal groundwork for a locally deployed model—likely aimed at security-sensitive customers who won’t touch cloud APIs.

What “Licensed Materials” Actually Covers

OpenAI’s new definition is specific: “Licensed Materials” are software, packages, code, containers, or other modules delivered “on local machines, private cloud infrastructure, or other customer-managed systems.” The license is limited, non-exclusive, non-transferable, and non-sublicensable. No modification. No redistribution.

For a studio that wants to fine-tune or embed an OpenAI model in a proprietary pipeline, that “no modification” restriction is the first red flag. You get what they ship, nothing more.

The Deletion-on-Termination Hammer

The killer line: “On termination, the customer must permanently delete the Licensed Materials and all copies thereof.” That’s not just the primary installation—it’s every backup, every cached container image, every copy that landed on a dev’s laptop for testing. Exit cost is spelled out in plain English.

Any business building a pipeline around on-prem inference needs to plan for that shutdown scenario from day one. Data migration, model replacement, fallback APIs—none of it matters if you can’t prove you wiped everything.

Why This Matters for Your Next Procurement

This is the kind of clause that signals product direction before any announcement. OpenAI is telling the market: we will sell you a model you can run inside your VPC, but we control the license, you don’t touch the code, and when you leave, you leave empty-handed.

Every enterprise architect evaluating local deployment for regulated or sensitive workloads will now have to weigh the convenience of an OpenAI model against the cost of a locked-in, zero-footprint exit. The terms are live today—read them before you build your roadmap.

Source: OpenAI Prepping for On-Prem Product?
Domain: ledger.somantix.ai