Improve Services
Definition: A catch-all justification — “to improve our services/products” — that Privacy Policies and Terms of Service use to cover a wide range of data-processing purposes (analytics, feature development, AI/model training, ad measurement) under one umbrella phrase. Risk level: This is one of the vaguest justifications a platform can give for processing personal data: a user has no way to know what concrete uses fall under “improving services,” and platforms tracked here use the same phrase for both low-risk purposes (fixing bugs) and higher-risk ones (training AI models).
Platforms Using This Clause
| Platform | Document Type | Date | Link |
|---|---|---|---|
| Privacy Policy | 2022-07-11 | link | |
| Privacy Policy | 2022-12-16 | link (same Meta-wide framing as Facebook’s) | |
| Terms of Service | 2022-07-11 | link | |
| Privacy Policy | 2025-06-10 | link | |
| Privacy Policy | 2025-05-01 | link (data-retention-period disclosure, not a training clause — see below) | |
| Terms of Service | 2023-08-02 | link (broad service-change justification, not AI-specific — see below) |
Common Wording
- In our interest to provide accurate and reliable reporting to our advertisers, developers and other Partners, to ensure accurate pricing and statistics on performance and to demonstrate the value that our Partners realise using Meta Company Products… — Facebook, Privacy Policy, 2022-07-11 (present in substance throughout) — ties “improving services” to advertiser/Partner reporting and analytics, one of several distinct purposes bundled under this catch-all in Meta’s “Data we use and why” table.
- Interactions with Meta’s [artificial intelligence technology] and related metadata. For example, messages you or others receive from, share with or send to Meta’s [artificial intelligence technology]. * Conduct research and analytics to evaluate the impact of new products and features… * In our interest to create, provide, support, improve and maintain Meta Products… — Instagram, Privacy Policy, 2022-12-16 (present in substance throughout) — the same Meta-wide “Data we use and why” table structure as Facebook’s, here paired explicitly with Meta’s generative-AI feature interactions as one of the data categories covered.
A big part of our Service is creating and using cutting-edge technologies that help us personalize, protect, and improve our Service on an incredibly large scale for a broad global community. Technologies like artificial intelligence and machine learning give us the power to apply complex processes across our Service. — Instagram, Terms of Service, 2022-07-11 (present in substance throughout) — explicitly names AI/machine learning as the “improve our Service” mechanism, rather than leaving it as an unspecified catch-all.
To train, fine-tune, evaluate and improve our Generative AI models used to create content (textual, audio, visual, and other media, or multimedia) on the LinkedIn platform (or elsewhere) and in LinkedIn’s lines of business. — LinkedIn, Privacy Policy, 2025-06-10 (present in substance throughout) — the most concrete confirmed use of “improve” tied directly to generative-AI model training, naming specific products (InBart, Collaborative Articles, Account IQ) that result from the data use.
The storage periods are determined on a case-by-case basis depending on the following factors: * The nature of the information and why it is collected and processed: The length of time we will keep your information will generally be determined by how long we need that information to provide you with our Services, including any optional features you use and to provide customer support. — Pinterest, Privacy Policy, 2025-05-01 — a different use entirely from every other platform tracked under this keyword: not a justification for data processing, but for data retention — broad, fact-specific language giving Pinterest wide discretion over how long data is kept, with no fixed deadline disclosed.
We work every day to improve our service, and we hope you love it. We may make changes to Pinterest from time to time and for various reasons - including adding, removing or changing features and how our services work. — Pinterest, Terms of Service, 2023-08-02 (present in substance throughout) — a plain, generic open-ended-change justification, not tied to AI or training at all, paired with an explicit “services like Pinterest aren’t perfect” disclaimer.
Notes & Trends
Facebook’s and Instagram’s Privacy Policy entries come from the identical Meta-wide “Data we use and why” table, confirming this is shared Meta infrastructure, not platform-specific drafting. Instagram’s Terms of Service is the clearest example of the catch-all made concrete: rather than leaving “improve our Service” vague, it explicitly names “artificial intelligence and machine learning” as the mechanism. LinkedIn’s Privacy Policy goes furthest of all four: its “improve” language sits directly inside the same sentence as “train, fine-tune, evaluate,” collapsing the vague catch-all and the concrete AI-training purpose into one clause — the clearest evidence in this wiki that “improve services” language can be a euphemism for AI-model training specifically, not just a separate, milder purpose alongside it. Pinterest’s Privacy Policy breaks the pattern entirely: its captured “improve services” text is a data-retention justification, not a data-processing one — a reminder that this keyword’s tagged text can describe a structurally different privacy-policy concept depending on the platform. Pinterest’s Terms of Service, by contrast, uses the keyword in its most generic sense: a plain “we make changes from time to time” disclaimer with no AI or data-rights framing at all.