High Level Claw (gohighclaw.app)
Effective date: 2026-09-18
Last updated: 2026-10-05
Operated by: Open Claw Assistance LLC, a Wyoming limited liability company
This notice is part of the Terms of Service. It explains how AI is used in the Service, what you must do before relying on it, and the disclosure duties that apply when the AI talks to your customers. Capitalized terms have the meanings given in the Terms of Service. "Fair use" in this notice means fair, lawful, and responsible use of AI features and of other people's content; it is not a claim that any particular use qualifies as fair use under copyright law.
1. Where AI is used
AI features in the Service include: the onboarding scan and Business Blueprint; website copy, structure, and images; ad scripts, storyboards, images, video, and voiceover; social captions and covers; reply drafts in the Inbox; Playbook and Automation text; the conversational Texting brain; the AI Receptionist that answers and places calls; the assistant ("Claw") that runs tools on your instructions; contract drafting; lead scoring; long-form video repurposing; and the My Video Clone avatar. Output is produced by third-party models through the provider you connected or, for platform-funded tasks, through our providers.
2. Output is a draft until a human approves it
2.1 What AI gets wrong. Output can contain factual errors, invented details ("hallucinations"), outdated information, bias, offensive material, text or images that resemble existing works, wrong prices, wrong hours, and claims your business cannot support. Models are non-deterministic: the same prompt can produce different Output.
2.2 Your review duty. You are responsible for reviewing Output before it is published, sent, signed, or acted on, and for everything that goes live from your Organization. The Service is built around this duty: every communication channel starts in review mode, previews are shown before renders, the assistant confirms before spending, sending, or deleting, and Playbooks created by the AI builder are saved disabled. Switching a channel to live is your decision and is recorded.
2.3 What to check every time. Facts about your business (name, hours, address, prices, services, licenses, guarantees); claims about results; comparisons with competitors; anything about health, safety, money, or the law; names and details of Contacts; the tone toward the recipient; and whether the Output includes something you did not intend to publish.
2.4 No guarantees. We do not guarantee that Output is accurate, complete, original, non-infringing, or fit for any purpose, and we do not guarantee any business result: leads, calls, bookings, sales, rankings, reach, engagement, or savings. Any estimate, projection, or benchmark shown in the Service is an illustration, not a promise.
3. AI-generated content labels
3.1 Default label. Social posts and captions produced by autopilot carry an "AI generated" label by default. Website images generated by the Service carry AI-generation metadata where the model provider supports it. If you remove a label, or edit content to remove it, compliance with any disclosure obligation is entirely yours.
3.2 Platform rules. Social platforms require disclosure of realistic AI-generated or altered media (for example Meta, TikTok, and YouTube). Ad platforms have their own rules for synthetic media in political and social-issue ads. Follow the rules of each platform you publish to.
3.3 State laws. A growing number of states regulate synthetic media: deceptive deepfakes in elections, non-consensual intimate imagery, and, in some states, disclosure when generative AI produces content that interacts with the public. Know the rules of the states where you operate and where your audience is.
3.4 Reviews and endorsements. Under the FTC rule on consumer reviews and testimonials, AI-written reviews, testimonials from people who did not use your service, and undisclosed incentives are illegal. Use AI to ask for real reviews, never to write them.
4. The AI Receptionist and outbound AI calls
4.1 Tell callers they are talking to an AI
The AI Receptionist introduces itself by default with a greeting of this form:
"Thanks for calling [BUSINESS NAME]. This is [PERSONA NAME], an AI assistant. This call may be recorded. How can I help you?"
Keep the AI disclosure. Several laws require or strongly favor it: California's Bolstering Online Transparency Act prohibits using a bot to communicate with a person in California to incentivize a purchase or influence a vote without clear disclosure; Utah's Artificial Intelligence Policy Act requires disclosure when a person asks whether they are talking to an AI and in regulated occupations; Colorado's AI Act imposes disclosure duties for consumer-facing AI systems; and the FTC treats undisclosed AI impersonation of a human as a deceptive practice. The Receptionist always answers truthfully if a caller asks whether it is a human.
4.2 Recording disclosure and consent
The Receptionist records and transcribes calls so you receive call records and summaries. Recording requires the consent that the law of the caller's location and your location requires.
- All-party consent states. California, Delaware, Florida, Illinois, Maryland, Massachusetts, Michigan (for third-party recording), Montana, Nevada, New Hampshire, Oregon (for telephone calls after recent amendments), Pennsylvania, and Washington require every party to consent to recording a phone call. Connecticut requires all-party consent for civil purposes.
- One-party consent states. The remaining states and federal law require the consent of only one party, which the business provides.
- Cross-state calls. Because a caller's location is often unknown, the Service treats every call as an all-party-consent call and plays the recording disclosure at the start. Continuing the call after the disclosure is the caller's consent in the all-party states listed above, except where a state requires an affirmative response; the Receptionist can be configured to ask "Is that okay?" and to stop recording if the caller says no.
- If you turn the disclosure off. Recording without the disclosure in an all-party-consent state can be a crime and can expose you to statutory damages. If you disable the disclosure or recording consent prompt in Receptionist settings, you accept that risk entirely, and you must have another lawful basis for recording every call.
- Callers who decline. When a caller declines recording, the Receptionist continues without recording, and you receive a summary generated from the live conversation only where that is lawful in the caller's state; otherwise you receive a notification that a call occurred.
4.3 Outbound calls placed by AI
Any call placed by the Receptionist or by an Automation uses an artificial voice. Under the TCPA, as the FCC confirmed in 2024, AI-generated voices are "artificial or prerecorded" voices. That means:
- Marketing calls to any mobile number, and prerecorded or artificial-voice marketing calls to residential lines, require the called party's prior express written consent that specifically covers artificial-voice or automated calls.
- Informational calls (appointment reminders, "we are running late", callback to a missed call) require prior express consent, which a customer gives by providing their number for that purpose.
- Every outbound AI call must state at the beginning that it is from your business, disclose that it is an automated or AI call, provide a callback number, and offer an opt-out mechanism (the Receptionist honors "stop calling me" by marking the Contact do-not-call).
- Hours. Outbound calls are limited to 8:00 a.m. to 9:00 p.m. in the recipient's local time, or the narrower window the state requires.
The Acceptable Use Policy contains the full communication rules. The Service's do-not-call marks, quiet hours, and consent ledger help you comply; they do not replace your obligation to obtain and keep proof of consent.
4.4 What the Receptionist will not do
The Receptionist is configured not to: give medical, legal, or financial advice; make promises about outcomes; quote prices not in your Business Blueprint; handle emergencies (it directs callers to 911 or local emergency services); collect payment card numbers, Social Security numbers, or passwords; or claim to be human. Configure its persona, scripts, and tools to match what your business can actually deliver, and review call summaries regularly.
5. Fair and lawful use of content and AI
5.1 Inputs you have rights to. Submit only content you own or are licensed to use: your photos, your videos, your documents, your brand assets, and information about your own business. Public availability is not a license. Repurposing a video found through Video Finder, using a competitor's photos from their website, or cloning a voice from a public recording is not fair use and is prohibited.
5.2 Likeness and voice. My Video Clone and voice features may be used only for your own likeness and voice, or for a person who has given you written consent that covers AI reproduction. Keep the consent.
5.3 Trademarks and copying. Do not use AI features to imitate a competitor's brand, copy a competitor's website, or create content designed to be confused with another business.
5.4 Copyright in Output. Under current United States Copyright Office guidance, content generated by AI without sufficient human authorship is not protected by copyright. Your edits, selection, and arrangement may be protectable. Do not assume you can stop others from using Output that closely resembles yours, and do not assume Output is free of third-party rights.
5.5 Provider policies. AI provider usage policies flow down to you through the BYOK and Third-Party Services Terms. In general they prohibit illegal uses, harassment, deception, malware, child sexual abuse material, non-consensual intimate imagery, weapons development, and uses that violate the rights of others.
5.6 Regulated advice and decisions. Do not use AI features to make or communicate decisions about housing, credit, employment, insurance, healthcare, or education for individuals, and do not let them give regulated advice unless you hold the license and you review the content.
6. How the AI uses your data
Prompts are built from your Business Blueprint, the Contact's messages or the call audio, and the document or asset in question. They go to the provider you connected, or to our providers for platform-funded tasks. We do not use your data to train models. Provider data-use terms and settings vary by API account and tier; a consumer subscription or a working API key does not establish no-training protection.
Calendar-derived available times, booking results and saved scheduling history require additional protection. AI scheduling may use Google Gemini, OpenRouter and its specifically reviewed model hosts, and optional ElevenLabs speech only after the destination's applicable no-training conditions and separate sharing consent have been established. Unknown eligibility does not authorize a fallback to another model or speech provider. Manual booking remains available. These requirements apply equally to Customer Keys and platform-funded processing. See Section 4 of the Privacy Policy for Google-data use, restrictions, disconnection and deletion.
7. Reporting a problem with AI Output
If Output produced by the Service is harmful, defamatory, infringing, or dangerously wrong, email support@openclawassistance.com with the Organization, the feature, and the Output id from the receipt. We investigate, adjust routing or safeguards, and, where a provider's model is at fault, report it to the provider.
8. Changes
We update this notice as AI laws, platform rules, and the Service change. Material changes are announced in the Service or by email.