Privacy Policy

1. Who we are and what this covers

pk2sl.un limited is a UK-registered, founder-led AI venture studio and Web 3.0 semantic, hybrid agentic communications infrastructure company. Company number: 17446976. Registered office: 71-75 Shelton Street, Covent Garden, London, United Kingdom, WC2H 9JQ. We are responsible for personal information used to run our own services. Contact jakub.kamecki@gmail.com for privacy, deletion or support requests.

This policy covers pk2sl.com, our ACAH (also called AHAC) assistant, and Pukot, Sulog and related experimental services where this policy is linked. A project-specific notice may add detail. Where we process information only on a business customer's instructions, that customer's notice and our processing agreement govern that activity.

2. Experimental services and active model training

We develop and test AI systems, human-judgment methods, evaluations, forecasting tools and career or business-support services. We actively train and evaluate models. These are experimental activities. Outputs, inferred assessments and research findings can be wrong, incomplete or change over time.

Depending on the stated project, information used for development may include authorised task contributions, feedback, evaluation examples and recorded outcomes. Personal information is used for training only where we have a lawful basis and the necessary rights; a separate consent or project notice is used where required. Visiting this website is not blanket consent to training, and submitting information does not override confidentiality, third-party rights or platform restrictions. We seek to minimise personal information and remove identifiers where practical. Pseudonymised information can still be personal data.

3. Information and sources

We receive information you provide through forms, email, conversations with our services, tasks and research: contact details, organisation or career information, goals, preferences, task responses, forecasts, feedback and outcomes. A paid activity may also require transaction and payment-administration records. Provide only information needed for the activity.

Services and hosting providers may process IP addresses, browser/device details, requests, timestamps, errors and security or usage events. An integration provides information needed for the feature you invoke. Where relevant, we may receive project information from a customer, contributor or publicly available professional source; an applicable project notice should identify additional sources and purposes. Do not submit passwords, API tokens, health information, financial account details or other sensitive or confidential information through general forms or bot commands.

4. ACAH / AHAC on Discord

The current Discord assistant responds to explicit commands. It does not continuously read server conversations. Discord sends interaction information, which can include your user, application, server and channel identifiers, the command and text you submit, timestamps, and a temporary interaction token used to respond. We use identifiers to check access and deliver the response.

The backend stores operational receipts, interaction identifiers, command names, status or error information and a pseudonymous user reference. A submitted Pukot question becomes a task and may be seen by assigned reviewers or contributors. A Sulog signal is stored for review. Private Discord responses do not mean the submitted information is never stored or reviewed.

Discord API data is not a general-purpose training corpus. We do not use API message content to train models without Discord's express permission. Discord API data must not be sold, used to profile users or determine employment eligibility, or repurposed beyond permitted app functions. Any separately offered research or career intake must follow its own notice and applicable permissions. Discord separately processes information under its privacy policy.

5. Why we process information

We process information to provide the feature you request, communicate with you, administer projects and contributions, assess task results, maintain records, prevent abuse, diagnose failures and meet legal obligations. Development and research may include evaluating or training models using appropriately authorised data.

Where UK GDPR or EU GDPR applies, our bases are: contract or requested pre-contract steps for delivering agreed services; legitimate interests for proportionate security, administration, service improvement and appropriately safeguarded research, subject to your rights; consent for activities that require it, including consent-based research or optional tracking; and legal obligation for required records. The applicable project notice explains any additional or more specific processing. We do not treat all of these bases as interchangeable.

6. Assessments and decisions

Pukot may assess response usefulness, consistency or calibration; Sulog may use information you provide to suggest career options. Experimental scores describe a task and context, not a diagnosis or a universal measure of a person. They may affect task invitations or project rewards where the programme explains this. Ask us to correct information or review an assessment. These tools are not a substitute for a responsible human decision, and this policy does not authorise solely automated decisions with legal or similarly significant effects. Discord API data is subject to the stricter limits above.

7. Access, providers and international processing

The founder and authorised people working on an activity may access information necessary for their role. Providers may process it for hosting (including Vercel), databases and backend services (including Supabase), communications, AI processing, document management, analytics or payments. Human reviewers may receive the task material needed for a requested review. We may disclose information where lawfully required, to protect rights or security, or in a business transfer with appropriate safeguards.

AI and infrastructure services may process information outside the UK or your home country. Where restricted international transfers occur, applicable transfer rules and safeguards must be met. Contact us for information about the providers and transfer arrangements relevant to your activity. We do not promise that every service or dataset stays in one country. Project-specific confidentiality or data-processing terms take precedence where applicable.

8. Retention, removal and trained models

We keep information for as long as needed for the activity, resolving requests or disputes, security, research reproducibility and records we are legally required to retain. The relevant criteria include whether a project is active, whether identifiable data remains necessary, contractual or statutory requirements, and whether information can be deleted or de-identified. There is no single retention period across all experiments.

You can request deletion at jakub.kamecki@gmail.com. Tell us the service and account or interaction concerned; do not send a password. We may need proportionate identity verification. We will act on valid requests under applicable law, remove data from active systems and future training inputs where required, and take reasonable steps with relevant processors. We will explain any lawful exception or practical limitation.

Deletion from source records does not necessarily reverse the influence of data in a model already trained. We cannot promise that every model can be individually “untrained”, or that a backup disappears immediately. We will assess what action the law requires, including measures concerning retained models, outputs or backups, and explain the response. These limitations do not waive your rights. Removing a bot from a server stops future use there; it is not automatically a deletion request for existing records.

9. Your rights and complaints

Depending on applicable law, you may request access, correction, erasure, restriction, portability, or information about automated processing. You may withdraw consent for future consent-based processing without affecting earlier lawful use.

Right to object: you may object to processing based on legitimate interests, including relevant research or model-development use, and to direct marketing. Contact us using the address above. We assess requests under the applicable law and normally respond to UK GDPR rights requests within one month, explaining any lawful extension.

You may complain to the UK Information Commissioner's Office or the relevant authority where you live. You need not contact us first.

10. Cookies, security and age

This site uses the Metricool visit tracker for aggregate audience measurement. According to Metricool, this tracker does not use cookies, does not store information on a visitor's device, does not assign unique visitor identifiers, and does not retain the visitor's IP address. It records aggregate visit and page-view information for site measurement. Hosting and linked services may also use technical features needed to serve or protect the site. Any future optional tracking that requires consent will be explained and offered through the relevant consent controls before it is enabled.

We use access restrictions and other reasonable safeguards, but no experimental or online system is risk-free. Do not use general intake for highly sensitive information. Participation in our experimental contributor and bot programmes is intended for adults aged 18 or older. Contact us if you believe a child has provided information.

11. Changes

We update this notice as services and data uses change. The date and version below identify this release. Material new uses may require further notice or consent before they begin. A policy update is not retroactive permission to use previously collected information for an incompatible purpose.