Comparisons

COMPARISON

Locke vs Enterprise AI Plans

Enterprise AI plans come with real contractual upgrades: no training on your data, retention controls, SOC 2 reports, region pinning, admin controls. They meaningfully lower the risk of using a single provider. What they don't change: your prompt still leaves the device, the provider can decrypt it, and provider staff can see it under their own access rules.

When Enterprise plans of major AI providers fits

  • Your organisation has standardised on one or two AI providers and signed enterprise contracts.
  • You need provider-level features (custom models, enterprise SSO, admin audit logs) that only the provider can offer.
  • You are comfortable that provider contractual controls and independent audits are sufficient for the data you send.

When Locke fits

  • Even with enterprise contracts, some categories of data should not leave your machine — privileged, PHI, customer PII, regulated financial data.
  • Your users mix sanctioned enterprise AI with unsanctioned tools and you want one consistent control.
  • You want defence-in-depth: enterprise contract on one side, on-device control on the other.

Side-by-side

DimensionLockeEnterprise AI plans
Where data goesSensitive fields never leave the device.All prompt content leaves the device, governed by contract.
Trust boundaryYour machine.The provider's infrastructure plus your contract.
Coverage of unsanctioned AISame controls regardless of which AI is open.None — applies only to the provider you contracted with.
Visibility for the providerProvider sees masked or tokenized data only.Provider sees raw prompt content.
Best forCategorically sensitive data and unsanctioned tool use.Standardising on a vetted provider for general workloads.

Bottom line

Enterprise AI plans reduce risk on the provider's side. Locke reduces it on yours. Most regulated teams end up wanting both — the contract for general use, the on-device layer for the data that should never leave the machine.