- 3.1.
The User must use the Service in accordance with applicable laws and regulations, good practice, and the terms applicable to the Service from time to time. The Service must not be used in a way that jeopardises its operation, infringes the rights of the Service Provider or third parties, or causes other harm. The User is responsible for their own device environment and for ensuring that the User’s use of the Service does not impede or disrupt the provision, maintenance, or use of the Service.
- 3.2.
The Service may include functionality enabling information sharing between Users, such as a commenting feature, for which the Service Provider may impose separate terms.
- 3.3.
If the User adds content to the Service, for example using the commenting feature, the User must ensure that they have the right to add (or transmit) the content to the Service and that the content does not infringe a third party’s intellectual property rights. The User grants the Service Provider an irrevocable, worldwide, royalty-free right to use content that the User adds or transmits to the Service. This right includes the Service Provider’s right to publish the content in the Service, use the content for purposes it considers appropriate (such as marketing the Service), and modify the content where necessary. This right applies only to content that the User publishes through the sharing or commenting features of the Service. It does not apply to research inputs, model tool calls, or outputs produced by the Service, which are governed by section 6.
- 5.1.
The Service Provider reserves the right, at its discretion, to change or remove the content, structure, parts, features, functionality, or availability of the Service at any time (including requiring registration as a condition for using the Service). The Service Provider may also, at its option, suspend or discontinue the provision and maintenance of the Service in its entirety.
- 5.2.
Unless otherwise required by mandatory law, the Service Provider does not guarantee that the Service will operate without interruption or error, nor is it responsible for information-security risks associated with its use or for any possible loss of data associated with the foregoing. This does not limit the Service Provider’s commitments under section 6.
- 5.3.
The Service is provided as is, and the Service Provider gives no warranties in relation to the Service or its content. The Service Provider is not responsible for the accuracy or timeliness of information presented in the Service or its fitness for a particular purpose.
- 5.4.
The Service Provider has the right to moderate, edit, and remove from the Service content that is contrary to the terms applicable to the Service and, for a justified reason, to prevent the User from accessing the Service (including by closing the User’s account), for example where the User uses the Service in breach of these Terms.
- 5.5.
Unless otherwise required by mandatory law, the Service Provider is not liable for any indirect damage, cost, or loss related to the Service, its content, or the use of either. The limitation of liability does not apply to damage caused intentionally or through gross negligence by the Service Provider.
- 5.6.
The Service Provider has the right to transfer the Service and its maintenance, as well as the rights, obligations, agreements, information, and materials related to the Service, to a third party.
- 6.1.
In connection with the Service, the Service Provider may collect and process Users’ personal data. Further information on the processing of personal data is available in the
Laki.ai Privacy Policy.
- 6.2.
When an authenticated User uses the research functions of the Service through any supported channel, such as ChatGPT, Claude, or another MCP connection, and the Service Provider processes personal data contained in a research input or output on the User’s behalf, the Laki.ai
Data Processing Addendum (DPA) is incorporated by reference into these Terms, and the User is then the Customer referred to in the DPA. This also applies to free accounts.
- 6.3.
A research input or tool call generated by the model must not contain PCI-regulated payment-card data, protected health information (PHI), government-issued personal identifiers, authentication credentials, or other secrets. Other regulated sensitive data, including special categories of personal data under Article 9 of the GDPR and personal data relating to criminal convictions and offences under Article 10 of the GDPR, may be processed only where strictly necessary for the legal task and where the User has ensured lawful control of the processing and legally adequate consent. Special categories of personal data and data relating to criminal convictions and offences may be provided only to the extent that they do not fall within the categories prohibited above; the User must remove personal identity codes, health information, and other prohibited data from the material before providing it. The User must minimise the input and ensure that they have the right to provide the data for processing by the selected host, Laki.ai, and the subprocessors.
- 6.4.
The Service Provider processes research inputs, meaning prompts, search queries, context, attachments, and other material processed by the research functions of the Service, including tool calls independently generated by the AI model, and outputs produced from them only for the agreed purpose of providing the Service; protects them with technical and organisational measures appropriate to the risk; does not use them to train or generally improve its own or third-party AI models; and retains them only for as long as the purpose of processing requires. Research inputs and research outputs produced by the Service are the User’s confidential information without separate marking, and the Service Provider uses them only to provide the Service. Research inputs may be processed temporarily and only to the extent needed to perform the requested search, document reading, model-based relevance assessment, or other operation, including in Google services (Google Cloud and Vertex AI). Model and embedding requests containing research inputs or research outputs produced by the Service are processed in the Vertex AI EU region. The Service Provider does not pre-screen or classify model tool-call content; its confidentiality and security obligations nevertheless apply to material it receives, and the same protection applies to research outputs produced by the Service. The Service Provider is responsible for its subprocessors under the DPA, and the DPA applies to research inputs and outputs to the extent they contain personal data. The content licence in section 3 does not apply to research inputs or outputs. The Service Provider may retain content-free metadata needed for security, operations, cost, and performance, such as user, session, tool and source identifiers, times, statuses, durations, counts, and token and cost data. The Service Provider does not use research-query or context text as product-development material.
- 6.5.
Personnel or a person acting on behalf of the Service Provider may access a research input or output only where technically possible and necessary for support requested by the User, development work separately agreed with the User, investigation of a serious abuse or security concern, a legal obligation, or a binding request from a public authority. Access is limited to what is necessary. The prohibition on training and general improvement in section 6.4 always applies and such access does not create an exception.
- 6.6.
A third-party AI assistant, agent, or host service selected by the User may process User material and model tool calls under its own terms, and the commitments in sections 6.4–6.5 do not apply to that processing. The Service Provider has no obligation to detect, classify, filter, or delete model tool-call content or prevent its transmission, and, subject to the limitations in section 5, no responsibility for the content selections or processing of the third-party service. The User must review the terms of use, privacy practices, service version, and settings of the selected service and ensure their right to transmit information to the entire processing chain.