Skip to content
    FollowUP Logo
    Back to Legal
    Schedule G

    AI Specific Terms

    Schedule G - EU AI Act compliance, deepfakes, AI disclosure obligations.

    FollowUP CRM - AI Specific Terms

    Schedule G to the Master Subscription Agreement

    Effective Date: 10 May 2026 Version: 2.0 Last Updated: 10 May 2026 Governing Language: English

    📋 Quick Summary (TL;DR)

    • What this is. Additional terms that apply when you use the AI features of the FollowUP CRM Platform - Conversation AI, Comment AI, Content AI, Workflow AI, AI voice agents, AI-generated content, and any other AI-powered feature.
    • Why this is a separate Schedule. AI law is changing fast. The EU AI Act phases in over 2025-2027. US state AI laws are emerging. By isolating AI terms, we can update AI rules without touching the rest of your contract.
    • Three big buckets. (1) What we provide - AI features, AI Sub-processors, safeguards. (2) What you must do - disclosure to end users that they're talking to AI, comply with EU AI Act Article 50, comply with US state AI laws, take responsibility for AI output. (3) What you must not do - EU AI Act Article 5 prohibited practices, high-risk uses without compliance, deepfakes of real people without consent, professional advice without a human in the loop.
    • AI training note. We do not, and we require our AI Sub-processors not to, use your inputs or your end-user data to train general-purpose AI models without your consent.
    • Disclaimers. AI output is not professional advice. AI features are provided "AS IS." You review, edit, and validate AI output before use. You indemnify us against AI-output claims you publish or transmit.

    1. Application of these AI Specific Terms

    1.1 When This Schedule Applies

    This Schedule G applies whenever you, your sub-account users, or your end clients use any AI Features of the Platform, including without limitation:

    • Conversation AI - automated chatbots and voice agents that interact via SMS, email, web chat, or voice;
    • Comment AI - AI-driven monitoring, classification, and reply generation for social media comments;
    • Content AI - AI-driven content generation (text, images, copy, marketing materials, social posts);
    • Workflow AI - AI-driven automation creation, decision-making, branching logic;
    • AI Voice Agents - synthetic voice systems that place or receive calls on your behalf;
    • AI-generated documents - proposals, estimates, contracts, summaries;
    • Premium Actions - AI-driven analysis, translation, sentiment, content transformation;
    • Decision Maker - multi-step AI-driven decision sequences.

    1.2 Relationship to Other Documents

    This Schedule G is incorporated into the Master Subscription Agreement (MSA) and supplements: MSA Section 6 (Communications and AI compliance for the sender role); AUP (Schedule B) Section 4 (AI-specific Acceptable Use rules); Privacy Policy; DPA (Schedule C).

    1.3 Automatic Incorporation

    By using any AI Feature, you accept these AI Specific Terms.

    1.4 Scope of Updates

    Because AI law is evolving rapidly, we may update this Schedule G more frequently. Material adverse changes require 30 days' prior notice; non-material changes posted with updated date. Where mandatory law requires immediate compliance, updates take effect when the law requires.

    2. AI Features - How They Work

    2.1 The General Architecture

    Our AI Features are powered by large language models (LLMs), generative AI systems, and other ML models operated by AI Sub-processors. Inputs are transmitted from the Platform to the AI Sub-processor; outputs are returned and delivered to your end users via configured channels.

    2.2 AI Sub-Processors

    Current AI Sub-processors include (without limitation):

    • OpenAI (USA) - GPT-class language models;
    • Anthropic (USA) - Claude-class language models;
    • Google AI / Google DeepMind (USA) - Gemini-class models;
    • Meta AI (USA) - Llama-class models, where used;
    • ElevenLabs (USA) - voice synthesis;
    • HighLevel native AI services;
    • other AI Sub-processors as listed on the Sub-Processor Page.

    2.3 Specific AI Sub-Processor Terms

    (a) OpenAI - API-tier inputs/outputs not used to train models by default. We rely on this commitment.

    (b) Anthropic - API inputs/outputs not used to train models without explicit opt-in (we do not opt in).

    (c) Google AI / Gemini API - enterprise/paid tiers that exclude data from model training where available.

    (d) ElevenLabs - voice synthesis processing subject to acceptable use policy, including restrictions on impersonation.

    (e) Meta AI - depends on access tier.

    2.4 Training Restriction - Our Commitment to You

    We do not use, and we contractually require our AI Sub-processors not to use, your inputs or your end-user data to train general-purpose AI models, except:

    (a) where the data has been irreversibly anonymized and aggregated; or (b) where you have expressly opted in to a specific feature or program in which AI training use is described in advance.

    2.5 Input Retention by AI Sub-Processors

    AI Sub-processors typically retain inputs for abuse monitoring purposes for short periods (commonly 30 days or less for OpenAI/Anthropic API tiers), then delete. For high-sensitivity uses, contact us to discuss enterprise tiers with shorter retention.

    2.6 Where AI Inputs Are Processed

    Typically located in the United States. Personal data flowing involves a Restricted Transfer under GDPR / UK GDPR / Georgian law, governed by the international transfer mechanisms in DPA Section 8.

    3. Your Obligations as a User of AI Features

    3.1 Review, Edit, Validate

    You are solely responsible for reviewing, editing, and validating all AI output before:

    (a) publishing externally; (b) transmitting to end users; (c) using as a basis for business decisions affecting natural persons; (d) representing as authoritative content (legal, medical, financial).

    You bear all risk of AI output.

    3.2 No Reliance on AI for Professional Advice

    You will not use AI Features to provide:

    (a) Individualized medical advice, diagnosis, or treatment recommendations; (b) Individualized legal advice ordinarily requiring a licensed attorney; (c) Individualized financial, investment, or tax advice; (d) Individualized therapy or psychological counseling to vulnerable individuals; (e) Determinations on credit, employment, insurance, education access, or public assistance constituting "automated decision-making with legal or similarly significant effects" under GDPR Article 22 without compliance.

    For general informational content in regulated areas, you may use AI Features provided there is a clear disclaimer, the AI directs to a qualified professional, and content is reviewed by a qualified professional.

    3.3 Disclosure That AI Is in Use - Your Primary Compliance Obligation

    When AI Features interact with natural persons on your behalf - chatbots, voice agents, AI sales reps, AI-driven email - you must disclose, clearly and at the start of the interaction, that the natural person is communicating with an AI system.

    Required under:

    (a) EU AI Act Article 50(1) - providers must ensure AI systems are designed so natural persons are informed they are interacting with an AI;

    (b) US state laws: California (SB 942, AB 1018, Bot Disclosure Law); Utah (AI Policy Act HB 149); Colorado (AI Act SB 24-205); Texas (Responsible AI Governance Act); Illinois (HB 3773); NYC Local Law 144;

    (c) FCC Declaratory Ruling (February 2024) - AI-generated voice calls fall within TCPA's prohibition on artificial or prerecorded voice calls without prior express written consent.

    3.4 Acceptable Disclosure Methods

    ChannelAcceptable Disclosure
    AI voice agentVerbal disclosure at start: "Hi, I'm an AI assistant calling on behalf of [your company]. This call may be recorded. To speak with a human, press 0 / say 'human' / hang up and call [number]."
    Chatbot (web/in-app)First message: "👋 Hi! You're chatting with an AI assistant from [your company]. I can help with [scope]. For complex matters, type 'human' to be connected to a person." Plus a persistent visual badge.
    AI-drafted SMS sent verbatim"(AI-generated message from [your company])" - or take responsibility as if you wrote it (with human review).
    Email autoresponder using AIDisclosure in body or signature: "This response was generated by our AI assistant. For human assistance, reply with 'agent' or contact [email]."
    AI-generated marketing contentDisclosure where required by FTC / EU guidance.

    Disclosures must be clear, at the start, in language the person can reasonably understand, and functional (option to talk to human must actually work).

    3.5 Exceptions to the Disclosure Obligation

    EU AI Act Article 50(1) allows an exception where it is "obvious from the circumstances and context of use" that the person is interacting with an AI. Do not rely on this exception lightly.

    3.6 Synthetic Content (Deepfakes)

    If you generate synthetic audio, video, or image content depicting real persons:

    (a) Consent. You must obtain the verifiable consent of any natural person whose likeness or voice is replicated.

    (b) Labelling - EU AI Act Article 50(4). You must label Synthetic Content as artificially generated or manipulated.

    (c) US state deepfake laws. Comply with applicable state laws including: Texas SB 751 (election deepfakes); California AB 730, AB 602, AB 2839; New York Civil Rights Law § 50-f, § 51; Tennessee ELVIS Act (voice replication of musicians); Illinois deepfake laws.

    (d) Absolute prohibitions. You must not generate Synthetic Content depicting real persons in: sexually explicit contexts (without explicit consent and lawful basis); defamatory contexts; deceptive identity-fraud contexts; electoral interference contexts.

    3.7 EU AI Act Article 5 - Prohibited AI Practices

    You may not use AI Features for any practice prohibited by EU AI Act Article 5, regardless of where you operate:

    (a) Subliminal or manipulative techniques beyond consciousness, that materially distort behavior in ways likely to cause significant harm; (b) Exploitation of vulnerabilities of natural persons due to age, disability, or social/economic situation; (c) Social scoring of natural persons by public authorities; (d) Predictive policing based solely on profiling or personality assessment; (e) Untargeted scraping of facial images from internet or CCTV; (f) Emotion inference in workplaces or educational institutions (except medical/safety reasons); (g) Biometric categorization to deduce protected categories; (h) Real-time remote biometric identification in publicly accessible spaces for law enforcement.

    Absolute and non-negotiable. Violation is grounds for immediate suspension.

    3.8 EU AI Act High-Risk Use Cases

    You may not use AI Features for any high-risk use case under EU AI Act Article 6 and Annex III without:

    (a) independently complying with all deployer obligations under EU AI Act Articles 26 et seq.: human oversight; data quality controls; log retention (typically 6 months); incident notification; DPIA where required; cooperation with authorities;

    (b) notifying us in writing at info@followup.agency of your intended high-risk use case.

    High-risk use cases under Annex III include: biometric identification; critical infrastructure; education and vocational training; employment and worker management; essential services and benefits; law enforcement; migration/asylum/border control; administration of justice; democratic processes.

    3.9 Sensitive Inputs

    You will not provide as inputs:

    (a) Protected Health Information (PHI) under HIPAA, unless you have a BAA; (b) Payment card data subject to PCI-DSS (full PAN, CVV, magnetic stripe data); (c) Government identification numbers (SSN, Tax ID, passport numbers) where avoidable; (d) Attorney-client privileged content that you intend to keep privileged; (e) Trade secrets of third parties without authorization; (f) Special-category Personal Data under GDPR Article 9 without lawful basis under Article 9(2); (g) Children's data subject to GDPR Article 8, COPPA; (h) Data subject to specific contractual confidentiality of third parties.

    3.10 Intellectual Property and AI Output

    You acknowledge:

    (a) AI output may be substantially similar to existing copyrighted/trademarked/protected material; (b) ownership of AI-generated content is legally uncertain (US Copyright Office: works without human creative authorship are not copyrightable); (c) you bear the risk of infringing third-party rights; (d) you indemnify us against IP and content claims arising from AI output you publish.

    3.11 Bias and Discrimination

    You must: monitor for biased output and remediate; not use AI Features to discriminate against protected groups; conduct bias testing for consequential decisions affecting natural persons; be prepared to explain AI-driven decisions.

    3.12 Logs and Records

    Maintain logs of AI Feature usage sufficient to: demonstrate compliance with this Schedule; investigate and remediate AI-output incidents; respond to regulator inquiries (EU AI Act Article 12 logging for high-risk uses); support data subject rights requests (GDPR Article 15(1)(h)).

    3.13 Children and AI

    Special care required where AI Features may interact with children - comply with GDPR Article 8, COPPA, applicable parental consent rules.

    4. Disclaimers and Limitation of Liability

    4.1 AI Features Are Provided "AS IS"

    AI Features are provided strictly "AS IS" and "AS AVAILABLE", without any warranty. We disclaim: any warranty as to accuracy, completeness, reliability, currency, or appropriateness; any warranty that AI output will be free of errors, biases, hallucinations, or infringement; any fitness for a particular purpose; any non-infringement warranty with respect to AI output content.

    4.2 No Liability for AI Output Consequences

    Subject to MSA Section 15, we are not liable for: consequences of your reliance on AI output; claims by third parties against you arising from AI output you transmitted; legal liability from failure to disclose AI use; AI Sub-processor outages or policy changes; loss of access to AI Features where Sub-processors withdraw services.

    4.3 Indemnification

    In addition to MSA Section 16, you specifically indemnify against: claims from your use in violation of this Schedule; claims by end users not adequately disclosed they were interacting with AI; IP infringement, defamation, privacy violation claims from AI output you used; regulatory action against your high-risk AI use case; claims arising from prohibited practices under EU AI Act Article 5 or US state equivalents.

    5. Specific AI Feature Notes

    5.1 AI Voice Agents

    (a) TCPA compliance. AI voice calls within the US subject to TCPA. Obtain TCPA-compliant prior express written consent.

    (b) STIR/SHAKEN. US carriers may require attestation. You are responsible for honest attestation.

    (c) Quiet hours. TCPA quiet hours (8 a.m. - 9 p.m. local time of recipient) and equivalent state rules.

    (d) Consent withdrawal. AI voice agents must recognize and immediately honor consent withdrawal ("STOP", "remove me", "do not call me again").

    5.2 Conversational AI Chatbots

    (a) Onboarding disclosure. First message in any session must disclose AI status.

    (b) Escalation to human. Offer an escalation path for complaint handling, refund requests, vulnerable user signals, complex queries.

    (c) Safety signals. Have safety protocols for users in crisis (self-harm signals, abuse signals, emergency situations).

    5.3 AI Content Generation

    (a) Plagiarism check. Run originality checks before publication in academic, journalism, formal publication.

    (b) Brand safety. Review for brand voice, tone, accuracy.

    (c) Cultural sensitivity. Review for cross-cultural audiences by humans with cultural fluency.

    5.4 AI in Workflow Automation

    (a) Decision points. AI-driven decisions should have clear logic, fallbacks, and override mechanisms.

    (b) Audit trails. Maintain logs of AI-driven decisions.

    (c) Article 22 compliance. Where AI decisions have legal/significant effects, ensure GDPR Article 22 compliance: meaningful information, human review, right to contest.

    6. Updates and Changes to AI Features

    6.1 Model Changes

    AI Sub-processors update models frequently. We endeavor to communicate material changes affecting Customer-facing functionality, but cannot guarantee notice of every model update.

    6.2 Feature Deprecation

    Where AI Sub-processors deprecate models, we may need to migrate. We will provide reasonable notice.

    6.3 Regulatory Adjustments

    As EU AI Act, US state laws, and other AI regulations evolve, we may add disclosure requirements; restrict certain uses; require documentation from Customers using high-risk applications; update AI Sub-processor selections.

    7. Withdrawal of AI Features

    We may withdraw or restrict AI Features where:

    (a) you have violated this Schedule, the AUP, or applicable AI law; (b) an AI Sub-processor has withdrawn services or imposed restrictions; (c) regulatory enforcement makes continued provision non-viable; (d) legal risk to us or HighLevel makes continued provision unreasonable.

    Withdrawal for cause does not entitle you to a refund. Withdrawal due to upstream changes may, at our discretion, entitle you to a pro-rata adjustment.

    8. Reseller Customers and AI Features

    If you are a Reseller Customer:

    (a) you must flow down the obligations in this Schedule to your end clients; (b) you are responsible for monitoring your end clients' AI use for compliance; (c) you must ensure your end clients' interactions include adequate AI disclosure; (d) violations by your end clients are treated as your violations.

    9. Contact

    For AI-specific matters:

    Sales Consulting Group LLC (operating as FollowUP CRM) Bakhtrioni 11, Tbilisi, Georgia Company ID: 405727254

    End of AI Specific Terms v2.0

    Other Legal Documents

    Terms of ServicePrivacy PolicyCookie PolicySub-ProcessorsAcceptable Use PolicyData Processing Addendum
    View all legal documents