AI Disclosure

SellerSlice LLC · Mission Control

TL;DR — the short version

Here's the honest, plain-language summary. The full document below says all of this in more detail, but if you only read one box, read this one:

• Where we use AI. Four places: writing and researching Amazon listing copy, scoring your listings for "Marketing-Ready," generating client reports, and an internal-only "Morning Review" news feed our team reads each morning.
• AI helps; humans decide. Every AI output is a draft or a recommendation. A person on our team — or you — reviews and approves it before anything happens to your Amazon account. No AI takes a legally significant action about you on its own.
• We do not train AI on your data. We do not use your Customer Data, and we never use Amazon-sourced data, to train, fine-tune, or improve any third-party AI model. Our agreements with AI vendors are being put in place to require the same of them — see the per-vendor status in §4 before relying on this for any specific vendor.
• Amazon data is walled off. Amazon-derived data is categorically excluded from all model training, benchmarking, and "service improvement." It is used only to serve your business, never pooled across clients.
• Who the AI vendors are. Anthropic and OpenAI (copy, scoring), xAI/Grok (internal news search only — never your data), and the Vercel AI Gateway (a routing layer). See the table in §4.
• You stay in control. You can ask us how a recommendation was made, ask a human to review it, and contest it. See §7 and §10.

Table of contents

1. Who this is from, and what this covers
2. Plain-language definitions
3. Which Mission Control features use AI
4. The AI vendors we use (per-vendor disclosure)
5. We do not train AI on your data — and we flow that down
6. Amazon data: categorically excluded from AI training and benchmarking
7. Human oversight: AI is advisory, people decide
8. AI input handling, retention, and security
9. International transfers of AI inputs
10. Your rights regarding automated processing (all in-scope jurisdictions)
11. Regulatory readiness — EU AI Act, AU, Quebec
12. Limitations: AI can be wrong
13. Contact us
14. Changelog and version history

1. Who this is from, and what this covers

1.1 This AI Disclosure is published by SellerSlice LLC, a Washington State (USA) limited liability company ("SellerSlice," "we," "us," "our"). It describes how artificial-intelligence features work inside Mission Control ("the Service") — our business-to-business internal-operations platform and client portal for Amazon-seller agency clients.

1.2 This document is part of, and should be read alongside, our Privacy Policy (v2.1), Terms of Service (v2.3), Sub-processors list, Data Processing Agreement, and Data Retention schedule. Where this document and the Privacy Policy describe the same thing (for example, AI vendors as sub-processors), they are written to say the same thing; if you spot a conflict, tell us at privacy@sellerslice.com and we'll fix the drift.

1.3 This disclosure is written for our launch jurisdictions: the United States (including California/CCPA-CPRA, Virginia, Colorado, Connecticut, Texas, and other state privacy laws), the United Kingdom, the European Union, Canada (including Quebec's Law 25), Mexico (LFPDPPP), and Australia (Privacy Act 1988 / Australian Privacy Principles). Jurisdiction-specific points appear in §10 and §11. Spanish-language Mexican users should read this together with our Spanish-language Aviso de Privacidad at /privacy#aviso-de-privacidad.

1.4 Mission Control is a business tool for adults (18+). It is not directed to consumers or children.

1.5 Lawful basis for AI processing (EU/UK). Where we process your personal data through AI features, we rely on the lawful bases set out in the Privacy Policy Article 6 basis matrix: performance of our contract with you for AI used to deliver the Service you've purchased (copywriting, Marketing-Ready scoring, and report generation done for your account); and legitimate interests for our internal operational features (the Morning Review news feed) and any product analytics, balanced against your rights and subject to your right to object (§10.1). We do not rely on AI processing for any purpose that would require separate consent under Article 6 without first obtaining it. We do not use special-category data for AI features.

2. Plain-language definitions

• AI / model. Software from a third-party vendor that generates text, scores content, or searches information in response to a prompt. We do not build or train our own foundation models.
• Prompt / input. The text and data we send to an AI model to get a result — for example, an existing listing's title and bullets sent to a copy model so it can suggest improvements.
• Output. What the model returns — a draft of copy, a score, a summary, a suggestion.
• Customer Data. Data you provide to, or that is generated for you within, the Service (as defined in the Terms). You own it.
• Amazon Information / Amazon-derived data. Any data exposed to us through Amazon's SP-API or Advertising API — catalog, listing content (titles, bullets, descriptions), orders, sales, settlement, financial, and advertising figures — and anything computed from it. Amazon's Data Protection Policy ("DPP") treats all of this as "Information" regardless of whether it is personal data.
• Advisory / decision-support. AI output that informs a human decision but does not itself take action. Everything our AI produces is advisory.
• Sub-processor. A vendor we use to help deliver the Service that may process data on our behalf. Our AI vendors are sub-processors; they are listed at /subprocessors.

3. Which Mission Control features use AI

Mission Control uses AI in four places. We are listing all of them — if we add a fifth, we'll update this section and the changelog (§14).

3.1 Copywriting and Copy Research (listing copy). When you or your account manager works on Amazon listing copy — titles, bullets, descriptions, A+ content, and copy variations — Mission Control can draft and research that copy with AI. The model may receive the relevant listing's existing copy and product context so its suggestions are on-brand and specific rather than generic. Because that existing copy is typically Amazon-sourced catalog content, this is one of the features where Amazon-derived listing content may transit an AI vendor (see §6 for how that content is restricted).

3.2 Marketing-Ready scoring. Mission Control scores listings against a 24-attribute readiness rubric ("Marketing-Ready"). Some of that scoring uses AI (including text embeddings) to evaluate copy quality and completeness. The listing copy it evaluates is also typically Amazon-sourced catalog content, so — like Copywriting (§3.1) — this is a path where Amazon-derived listing content may transit an AI vendor, and it is covered by the same §6 Amazon restrictions. The score is a recommendation to a human reviewer — it does not auto-publish, auto-reject, or change your Amazon listing on its own.

3.3 AI report generation. Mission Control can draft client-facing reports (narrative summaries of performance and activity). AI helps write the prose; a human reviews the report before it is shared with you. Because performance narratives may incorporate Amazon-sourced figures, this feature is also covered by the §6 Amazon restrictions.

3.4 Morning Review / News (internal only). Each morning, an internal feed surfaces relevant industry and account news for our team to triage. This feature uses web and X (formerly Twitter) search via xAI/Grok. It does not process your Customer Data or any Amazon-sourced data — it searches public sources for our internal operational awareness. You never see this feature; it is an employee tool.

4. The AI vendors we use (per-vendor disclosure)

We use four AI vendors/layers in production. We list each one separately — purpose, location, whether it touches Amazon data, and how inputs are retained — because a single "we use AI" line is not honest enough.

Substantiation gate (read first). The no-training and short-retention statements throughout this document are only as strong as the actual signed vendor terms. Until a vendor's no-training/retention tier and processing region are confirmed, we do not publish an unqualified no-training claim, a retention claim, or a transfer-mechanism representation for that vendor. The "Confirmation pending" entries in the table below mark exactly where confirmation is still pending. See §4.3.

4.1 These vendors also appear on our Sub-processors list, which carries a "last updated" date, a change-notice commitment, and a 30-day advance-notice and objection window for material changes.

4.2 Amazon is shown elsewhere in our documents as a data source and independent controller, not as our AI sub-processor. Amazon's SP-API and Advertising API are where Amazon Information originates; Amazon is not part of our AI inference path.

4.3 Substantiation gate (detail). Every no-training and short-retention statement in §5 and §8 is contingent on the vendor confirmations noted above. Where a vendor's terms do not yet guarantee no-training or short retention, we will narrow or withhold the claim for that specific vendor rather than make a blanket statement we cannot stand behind. No unqualified per-vendor no-training claim is published while that vendor's row above is marked "Confirmation pending."

5. We do not train AI on your data — and we flow that down

5.1 It is our standing instruction and contractual requirement that your Customer Data, and Amazon-sourced data, are not used to train, fine-tune, or improve any third-party AI model. Our agreements with AI sub-processors are being put in place to flow this prohibition down to each vendor — meaning the vendor is contractually required not to use the data we send it to train or improve its models. This flow-down is asserted as a confirmed, present-tense fact only for vendors whose contracts have been verified per §4.3; for any vendor still marked "Confirmation pending" on no-training in §4, treat this commitment as our requirement and design intent pending that confirmation, not as a settled vendor representation.

5.2 Where an AI feature must send content to a vendor to produce a result (for example, sending existing listing copy to draft new copy), that transmission is for inference only — the model reads the input, returns an output, and the input is not used to train the model. Inference inputs are retained, if at all, only for the short operational window the vendor's no-training tier permits, once that tier is confirmed (see §4 and §8).

5.3 This commitment is also written into our Terms of Service §7 (data license): you own your Customer Data; we take only a narrow operational license to run, secure, and support the Service; any de-identified or aggregated analytics we perform is limited to non-Amazon data; and all Amazon-derived data is categorically excluded from training, benchmarking, and service-improvement.

6. Amazon data: categorically excluded from AI training and benchmarking

Amazon data gets the strictest treatment in Mission Control, because Amazon's contracts (the Amazon Data Protection Policy, the Acceptable Use Policy, and the Solution Provider Agreement — controlling text effective November 25, 2025) require it.

6.1 Never used to train any model. Amazon Information is never used to train, fine-tune, or improve any AI, ML, or LLM model, ours or a vendor's. This is non-negotiable and is flowed down to our AI sub-processors. (Amazon DPP; Solution Provider Agreement Permitted-Use limitation, §D-7.1/§D-7.5.)

6.2 Never pooled or benchmarked across clients. Amazon Information is used solely to support the Authorized User (you) whose account it came from. It is never aggregated across clients, never used for cross-client benchmarking, and never disclosed to any other client — and Amazon's rules make clear that de-identifying or aggregating the data does not lift this bar. (Amazon AUP §4.4 and §4.6 — the §4.6 bar reaches data whether individually identified or aggregated; Solution Provider Agreement §D-7.1/§D-7.5.) If you've seen a "compare me to your other clients" benchmark elsewhere, we do not and will not build that from Amazon data.

6.3 No buyer personal data. We use no Restricted SP-API roles, so we hold no buyer personal information — no buyer names, addresses, emails, or phone numbers. AI inputs derived from Amazon therefore do not contain buyer PII.

6.4 Transient only, and deleted on disconnect. When an AI feature uses Amazon-derived listing content (§3.1, §3.2, §3.3), that content transits the vendor only for inference and is not retained for training. Separately, when you de-authorize, disconnect, or close your account, all Amazon Information we hold is deleted (destroyed to a NIST 800-88 standard) within 30 days — not anonymized. Anonymization is categorically not an accepted substitute for deletion under the Amazon DPP. See the Privacy Policy Amazon API data-handling section and the Data Retention schedule for the full deletion mechanics.

7. Human oversight: AI is advisory, people decide

7.1 Every AI feature in Mission Control is advisory / decision-support. AI drafts, scores, summarizes, and suggests. It does not take legally significant or similarly significant action about you on its own.

7.2 A human reviews before any account action. Before anything happens to your Amazon account, a person — your account manager, a copywriter, a reviewer, or you — reviews and approves the AI output. AI copy is editable before use; Marketing-Ready scores are reviewed by a person; AI-drafted reports are reviewed before they're shared.

7.3 Not a solely-automated decision (design intent). Because of this human-in-the-loop design, our AI features are designed not to constitute decisions "based solely on automated processing" that produce legal or similarly significant effects on an individual (GDPR Article 22; UK GDPR as amended by the Data (Use and Access) Act 2025; Quebec Law 25 art. 12.1). They are decision-support tools that inform human judgment.

7.4 You can ask, review, and contest. You can ask us how a recommendation was reached, request that a human review or re-review it, and contest the outcome. To do so, contact privacy@sellerslice.com or use the data-request form. See §10 for the jurisdiction-specific rights behind this.

7.5 Can AI features be turned off? To ask about disabling AI-assisted features (such as AI-assisted copywriting, Marketing-Ready scoring, and report drafting) for your account, or about the scope of any such opt-out, contact us at privacy@sellerslice.com.

8. AI input handling, retention, and security

8.1 What gets sent. Only the data necessary to produce the result is sent to an AI vendor — for example, the listing copy and product context needed to draft a better listing. We follow a data-minimization principle for AI inputs.

8.2 Retention of inputs. AI inputs are treated as transient: they are sent for inference and are not used for training. Any short-term retention by a vendor is governed by that vendor's no-training tier once confirmed (§4.3); see the per-vendor rows in §4 and the Data Retention schedule.

8.3 No buyer PII in inputs. As noted in §6.3, we hold no buyer PII, so AI inputs do not contain it.

8.4 Security. AI traffic rides the same security posture as the rest of the Service: encryption in transit (TLS 1.2+), encryption at rest for stored data, tenant isolation, and role-based access. The full technical and organizational measures (TOMs) are described in the technical-and-organizational-measures annex to our Data Processing Agreement.

9. International transfers of AI inputs

For any AI vendor whose processing region is still marked "Confirmation pending" in §4, no transfer representation is published until its region and the corresponding mechanism (SCC module / DPF certification status / UK Addendum) are verified for that vendor; the mechanism bullets below cannot be relied on for such a vendor.

When AI inputs travel to a vendor outside your country, we rely on the same transfer framework described in the Privacy Policy transfers section and the DPA. The applicable mechanism depends on the specific vendor and its confirmed processing region:

• EU → US: For vendors certified under the EU-US Data Privacy Framework (DPF) — such as Stripe and Vercel — the DPF certification is the primary transfer mechanism, with EU Standard Contractual Clauses as a fallback. For non-DPF-certified vendors — such as Supabase and Resend — EU Standard Contractual Clauses are the mechanism, supported by a documented transfer impact assessment. The mechanism for each AI model vendor (Anthropic, OpenAI, xAI) will be selected on the same basis once its region and certification status are confirmed per §4.
• UK → US: the UK International Data Transfer Agreement, or the International Data Transfer Addendum to the EU SCCs, as applicable to the vendor.
• Mexico → US: transfer on the basis of consent and contractual equivalence (no adequacy regime; we name US recipients in the Spanish-language Aviso de Privacidad at /privacy#aviso-de-privacidad, with an accept/reject choice for voluntary purposes).
• Canada/Quebec → US: a written agreement reflecting a privacy impact assessment under Quebec Law 25 before Quebec personal information lands on the US stack.
• Australia → US: APP 8 "reasonable steps" via enforceable contracts binding each US recipient to substantially-equivalent handling.

9.1 Vendor processing regions for Anthropic, OpenAI, and xAI/Grok are flagged for confirmation in §4. We will state the actual regions and the matching mechanism once confirmed, because the transfer mechanism depends on the destination.

10. Your rights regarding automated processing (all in-scope jurisdictions)

You have rights over how your personal data is processed, including by AI features. We honor these across every launch jurisdiction. The full rights catalog (access, correction, deletion, portability, and the rest) lives in the Privacy Policy data-subject-rights section; below are the parts most relevant to AI.

10.1 European Union & United Kingdom (GDPR Arts. 13–15, 21, 22; UK GDPR as amended by the DUAA 2025). You have the right to be informed about automated processing, to access information about it, to object to processing we carry out on the basis of legitimate interests (see the lawful-basis statement in §1.5), and — for decisions based solely on automated processing with legal or similarly significant effects — to obtain human intervention, express your view, and contest the decision. As explained in §7, our AI features are designed to be human-reviewed and not to fall into the solely-automated category; even so, you may request human review and contest any output.

10.2 United States — California (CCPA/CPRA) and Virginia / Colorado / Connecticut / Texas and other state laws. You have the right to know, access, correct, and delete personal information, to opt out of certain profiling, and — under the Virginia-model state laws — to an appeal if we deny a request. California uniquely covers business-to-business contacts and employees. We do not sell or share your personal information. We honor the Global Privacy Control (GPC) signal as an opt-out of the sale/share of personal information and of targeted advertising; that GPC opt-out applies at all times and cannot be re-enabled away. For this product specifically, we also treat GPC as a default opt-out of analytics — and analytics (only) may be affirmatively re-enabled by you in our cookie settings. To be clear: re-enabling analytics changes the analytics default only; it does not lift, weaken, or re-enable the sale/share or targeted-advertising opt-out, which GPC continues to bind regardless. The same carve-out is mirrored in /cookies. Your "Your US State Privacy Rights" details, including the appeal path, are in the Privacy Policy.

10.3 Canada — PIPEDA and Quebec Law 25. You have meaningful-consent rights (what data, with whom, why, and risk of harm — including that data is processed by US processors and stored in the US), data portability, de-indexing/erasure, and — under Quebec Law 25 art. 12.1 — the right to be informed of, and to contest, a decision based exclusively on automated processing, and to submit observations to a member of our staff. Our designated Person in charge of the protection of personal information (privacy officer) is named in §13.

10.4 Mexico — LFPDPPP (ARCO rights). You have rights of Access, Rectification, Cancellation, and Opposition. We respond within 20 business days and implement granted requests within 15 business days. AI/profiling and any analytics that survive Amazon's bar are placed in the voluntary (finalidades secundarias) bucket in the Spanish-language Aviso de Privacidad at /privacy#aviso-de-privacidad, which you can decline without losing the Service.

10.5 Australia — Privacy Act 1988 / APPs. You have rights of access and correction (APP 12/13). Where an AI feature substantially informs a decision significantly affecting you, our privacy notice will disclose that automated decision-making (see §11.3).

10.6 How to exercise these rights. Use the data-request form or email privacy@sellerslice.com. We offer at least two submission methods, support authorized agents, and provide an internal appeal path for denials. We respond within 30 days (up to 45 days for certain US states; the Mexico timers above apply for Mexican requests).

11. Regulatory readiness — EU AI Act, AU, Quebec

We are tracking AI-specific transparency obligations that take effect over the next several months and will update this disclosure and add any required in-product labeling on or before each deadline.

11.1 EU AI Act, Article 50 (transparency). For interactions with AI systems and for AI-generated or AI-assisted content, Article 50 transparency obligations are expected to apply (currently dated August 2, 2026). We will provide the required in-product AI labeling on the affected features before the applicable date.

11.2 Quebec Law 25, art. 12.1. As described in §10.3, where a decision is based exclusively on automated processing we will inform you and honor your right to contest. Our features are human-reviewed, but we surface this right proactively.

11.3 Australia — automated-decision-making disclosure (by December 10, 2026). If any AI feature comes to substantially inform a decision that significantly affects an individual, we will disclose that in our privacy policy by the statutory deadline. We currently assess our features as decision-support, not significant-effect automated decisions; we will re-assess as features evolve.

11.4 Re-assessment commitment. If we change how a feature works — for example, removing a human-review step or adding a new determinative use — we will re-assess these obligations before the change ships, update this disclosure, and note it in the changelog.

12. Limitations: AI can be wrong

12.1 AI output can be inaccurate, incomplete, or out of date. It is generated by third-party models and is provided as-is, without warranty, as set out in the Terms of Service.

12.2 Because AI is advisory (§7), a human should always verify AI output before relying on or acting on it. We do not warrant any sales, ranking, ACoS, Buy-Box, or other Amazon outcome from AI-assisted work. AI suggestions are a starting point for human judgment, not a guarantee.

12.3 You own the AI-assisted output produced for you (for example, listing copy), as set out in the Terms. We do not feed Service-derived data into external models for training (transient inference only).

13. Contact us

Quebec Law 25 — Person in charge of the protection of personal information (privacy officer): Jordache Perozzo, President — privacy@sellerslice.com.

Amazon security-incident point of contact (IMPOC): We maintain a named Amazon Information Security Point of Contact, reachable via privacy@sellerslice.com. We notify security@amazon.com within 24 hours of detecting a qualifying security incident.

Mailing address / Mexico privacy contact (Departamento de Datos Personales): SellerSlice LLC, 1625 SE 192nd Ave, Suite 210, Camas, WA 98607, USA — privacy@sellerslice.com.

14. Changelog and version history

We will post material changes to this disclosure with a new version number and effective date, and keep prior versions available. If a change materially affects how AI processes your data, we'll tell you in line with our Privacy Policy change-notice commitment.