Terms of Service
§ 1 Scope
These Terms of Service apply to all contracts between StartLobster (hereinafter "Provider") and the Client regarding the provision of services for the setup, configuration, and management of OpenClaw instances.
§ 2 Scope of Services
The Provider delivers the following services:
- Installation and configuration of OpenClaw on the Client's infrastructure
- Security hardening of the OpenClaw instance
- Integration with messaging platforms and business tools
- Ongoing maintenance and monitoring (under Managed Care agreements)
The scope of services expressly does NOT include:
- Creation, review, or optimization of prompts, instructions, or system configurations that control AI behavior
- Monitoring or moderation of AI-generated content
- Ensuring the accuracy, completeness, or appropriateness of AI outputs
- Protection against prompt injection, jailbreak attempts, or social engineering attacks on the AI
- Compliance with industry-specific regulations regarding the Client's AI usage
§ 3 Client Responsibilities
The Client is responsible for:
- Providing required API keys (e.g., Anthropic, OpenAI)
- Designing and maintaining system prompts and AI instructions
- Monitoring and moderating AI-generated content and actions
- GDPR compliance regarding personal data processed by the AI
- Secure handling of credentials and API keys
- Ensuring AI usage complies with applicable laws and regulations
- Regular review of AI outputs for accuracy and appropriateness
§ 4 Limitation of Liability
4.1 The Provider bears unlimited liability for damages arising from injury to life, body, or health, as well as for damages caused intentionally or through gross negligence.
4.2 In cases of slight negligence, the Provider is only liable for breaches of essential contractual obligations (cardinal duties). Liability is limited to foreseeable, contract-typical damages.
4.3 The Provider is expressly not liable for:
- Damages caused by incorrect, incomplete, or misleading AI outputs
- Damages caused by prompt injection, jailbreaking, or other third-party AI manipulation
- Data losses or leaks resulting from the Client's AI usage
- Damages resulting from the Client sharing confidential information with the AI
- Privacy law violations resulting from how the Client uses the AI
- Business decisions based on AI-generated information
- Reputational damages from autonomous AI actions not adequately supervised by the Client
4.4 Total liability for all claims under a contract is limited to the fees paid in the preceding 12 months.
§ 5 AI-Specific Notices
5.1 OpenClaw is an open-source AI assistant based on Large Language Models (LLMs). LLMs can produce erroneous, biased, or unpredictable outputs. The Client acknowledges this.
5.2 Known risks of AI assistants include:
- Prompt Injection: Third parties may attempt to manipulate the AI via crafted inputs (e.g., in emails, web pages)
- Hallucinations: The AI may generate plausible-sounding but factually incorrect information
- Data Leaks: Insufficient configuration can expose confidential data through AI outputs or connected platforms
- Autonomous Actions: The AI can execute shell commands, modify files, and send messages — with real consequences
5.3 The Provider strongly recommends:
- Regular review of AI outputs by the Client
- Restricting AI permissions to the minimum necessary
- Never inputting highly sensitive data (e.g., passwords, banking details) into AI conversations
- Implementing approval processes for critical AI actions
§ 6 Data Processing
Where the Provider gains access to Client personal data under Managed Care services, a separate Data Processing Agreement (DPA) pursuant to Art. 28 GDPR will be concluded.
§ 7 Contract Term and Termination
Managed Care agreements have a minimum term of 3 months and automatically renew monthly unless terminated with 30 days' notice before the end of the month.
§ 8 Final Provisions
German law applies. Place of jurisdiction is the Provider's registered office, to the extent permitted by law. If any provision of these Terms is invalid, the remaining provisions remain in effect.
Last updated: February 2026