# Compliance (/docs/compliance)



LinkFetch is built on three commitments:

1. **Public data only.** We never read anything you couldn't read by
   visiting a profile while signed in to LinkedIn.
2. **User as principal.** Profile, company, post, group, search, and
   outbound endpoints all run inside the **end user's own** signed-in
   LinkedIn session — through the
   [LinkFetch Chrome extension](/docs/extension). LinkFetch the
   service never logs in to LinkedIn on your behalf.
3. **Provenance by default.** Every row carries `source`,
   `fetched_at`, `freshness_days` so you can build agent workflows for
   regulated customers without an opaque-data audit hole. See
   [Provenance](/docs/provenance).

This is what hiQ v. LinkedIn settled into: scraping public-page data
on behalf of an authenticated user is fine; impersonating LinkedIn
with fake accounts is not.

## Data subject rights (DSR) [#data-subject-rights-dsr]

LinkFetch honours the European GDPR right to erasure (Art. 17), the
California CCPA right to delete, and the equivalent Brazilian LGPD
right.

### Self-service [#self-service]

Email **[privacy@linkfetch.io](mailto:privacy@linkfetch.io)** with the
profile URL or vanity slug to suppress. Within **30 days**:

* The profile is removed from the cache.
* The slug is added to a permanent suppression list — re-fetches
  return `410 gone`, including via the extension capture path.
* Any cached search results, post reactors, or company-employee rows
  referencing the profile are scrubbed.

### Programmatic [#programmatic]

Pro-tier customers and above can call the suppression endpoint
directly:

```bash
curl -X POST https://api.linkfetch.io/v1/compliance/suppress \
  -H "Authorization: Bearer sk_live_..." \
  -H "Content-Type: application/json" \
  --data '{"url":"linkedin.com/in/example","reason":"dsr"}'
```

We respond with the suppression hash and the date by which all
downstream caches will have purged. Suppression is **permanent and
irreversible** — design accordingly.

## What we don't do [#what-we-dont-do]

* We don't run automated headless LinkedIn accounts. Read endpoints
  resolve through the user's own session via the extension; the
  Jobs scraper only hits the public-guest `/jobs/view/<id>` paths.
* We don't sell data — we sell *typed access* to data that is already
  public to a logged-in viewer.
* We don't keep a "people database" you can browse outside the
  context of an explicit lookup. Profile lookups debit credits and are
  logged per request id, per API key.
* We don't store LinkedIn session cookies on our infrastructure. They
  stay in the browser.

## Data retention [#data-retention]

| Class                                     | Retention                                                    |
| ----------------------------------------- | ------------------------------------------------------------ |
| Cached profiles, companies, posts, groups | 30 days, then re-fetched on next access                      |
| Jobs dataset rows                         | Until the underlying job is closed by the employer + 90 days |
| Request logs (`request_id` + params)      | 90 days                                                      |
| Credit-ledger entries                     | 7 years (financial requirement)                              |
| Suppression list                          | Permanent (that's the point)                                 |

## DPA, SOC 2, sub-processors [#dpa-soc-2-sub-processors]

Pro and Scale tiers ship with a DPA on request. We are SOC 2 Type II
in progress (target completion 2026-Q3).

Sub-processors:

| Vendor          | Purpose                                   |
| --------------- | ----------------------------------------- |
| AWS (us-east-1) | Hosting, primary Postgres, object storage |
| Paddle          | Merchant of Record for billing            |
| Cloudflare      | DDoS / WAF on the API edge                |
| Resend          | Transactional email                       |

Email &#x2A;*[security@linkfetch.io](mailto:security@linkfetch.io)** for
the SOC 2 readiness packet, the DPA template, or the current
sub-processor list.

## Acceptable use [#acceptable-use]

The LinkFetch terms forbid:

* Using LinkFetch to build a scraper-style "lead list" decoupled from
  a user-authorised lookup.
* Sending unsolicited outbound DMs at scale (we already block strangers'
  DMs at the API level — no first-degree, no message).
* Impersonation, harassment, or any content that would violate
  LinkedIn's own community rules.

We will close accounts that violate these. Email
[info@linkfetch.io](mailto:info@linkfetch.io) before scaling outbound
volume so we can size the limits to your use case.