1. Current source status
Siftglow's Reddit and X source integrations are currently provider-pending. Siftglow will not collect live platform data until the relevant official access, commercial permission, credentials, and operating limits are approved and enabled. It will not use scraping, proxies, simulated logins, account rotation, or other unapproved access as a fallback.
2. Approved public content only
If a source is approved and enabled, Siftglow will use only the official access path and scope granted for that source. The product is designed for public posts and conversations, not direct messages, private communities, private profiles, private group content, modmail, social-account credentials, cookies, or other private-area data.
3. Minimum data
Siftglow proposes to retain only the fields needed for evidence-led review:
- the platform and provider item identifier;
- a public display name only where the approved platform terms permit it;
- the post time, exact permalink, and the minimum title or excerpt needed as evidence;
- the original content language or an honest unknown value, plus how that language value was obtained;
- retrieval time and an available, unavailable, or deleted state.
Siftglow does not need profile pages, avatars, karma, scores, follower graphs, email addresses, phone numbers, social credentials, or full raw API-response archives for this use case.
4. Source integrity and private data flow
If approved, the intended flow is: official source access → approved-field minimization → deduplication and Project-rule matching → a private authenticated review queue → the exact original permalink. Siftglow does not publish a public platform-data feed or sell the resulting records as a dataset.
Siftglow preserves the source platform, provider item identifier, timestamp, original language, and exact permalink needed to distinguish source facts from product inferences. It does not fabricate a source post, substitute an example for missing provider data, or present an inference as a statement made by the original author.
5. Human-controlled participation
Siftglow separates source facts, product inferences, unknowns, and suggested actions. It may help a customer prepare a draft, but it never posts, replies, votes, follows, messages, or sends a direct message on the customer's behalf. The customer opens the exact platform permalink and controls every external action on Reddit or X.
6. Prohibited uses
Siftglow platform data must not be used for:
- advertising audiences, data resale, identity graphs, or off-platform identity matching;
- re-identification, sensitive-trait inference, or surveillance of sensitive people, groups, or events;
- employment, housing, credit, insurance, education, or similar high-impact decisions;
- automated spam, harassment, deceptive outreach, or unsolicited mass communication;
- training, fine-tuning, or evaluating a general-purpose or foundation model; or
- bypassing platform permissions, access controls, rate limits, or removal requirements.
7. AI processing boundary
The currently enabled DeepSeek workflow processes product-page content submitted through the product URL flow to generate a product snapshot and keyword suggestions. It does not currently receive Reddit or X content. Siftglow will not send platform content to an external model provider unless the relevant platform expressly permits that processing and the named processor, retention, non-training, security, and deletion terms are documented.
8. Untrusted content and prompt injection
Public posts and comments are untrusted input. Text inside a source item is treated only as quoted evidence and cannot authorize a rule change, external connection, data disclosure, deletion, purchase, or external action. If future approved processing uses an AI system, source content must remain separated from trusted instructions, permissions must remain least-privileged, and a human must control any action outside Siftglow.
9. Retention, updates, and deletion
Identifying source evidence is limited to the applicable plan window described in the Privacy Policy and never beyond a stricter platform or contractual limit. Upstream deletion, removal, revocation, account deletion, and contractual deletion requests take priority over ordinary retention. When a required deletion signal is received, Siftglow is designed to stop use and display and remove the related permalink, author identifier, excerpt, and recoverable derived artifacts.
10. Independence and contact
Reddit and X are trademarks of their respective owners. Siftglow is an independent service and is not affiliated with, endorsed by, or sponsored by Reddit or X.
Questions, removal notices, and platform-data requests may be sent to support@siftglow.com.