AI Transparency and Responsible Use Policy

1. General Description of AI Systems

1.1 Seek — Career Discovery and Matching

Seek uses supervised machine learning models trained with labor market data, occupational taxonomies, and user career outcomes to generate career path recommendations. The models incorporate natural language processing to analyze resumes and free text input, and collaborative filtering to identify career patterns in comparable user profiles. Seek models are periodically retrained with updated labor market data sets to keep recommendations relevant.

1.2 TalentOS — Talent Intelligence and Orientation

TalentOS uses machine learning classifiers and regression models to generate talent readiness scores, skills gap analysis, and institutional talent reports. The models are trained with validated competency frameworks aligned with recognized standards, including O*NET, the European Digital Competence Framework (e-CF) and datasets from institutional partners shared under applicable data agreements. TalentOS scores are designed to complement qualified human professional judgment, not replace it.

1.3 PathMaker — Autonomous Content Operations

PathMaker integrates large language models (LLMs) using API access to generate professional content, including resumes, cover letters, LinkedIn summaries, articles, and career communications. Content generation is configured through prompt engineering for professional precision and is subject to layers of content filtering and toxicity detection. PathMaker does not generate content designed to mislead employers, falsify credentials, or misrepresent a user's professional history.

2. Principles of Transparency in AI

• Explainability: Where technically possible, SIKER provides users with natural language explanations of the main factors that influence career recommendations and talent scores. Explanations internal to the application are available at the result level.
• Disclosure: SIKER will always indicate when content or assessments have been generated or substantially influenced by AI systems. AI-generated results are labeled within the platform interface.
• Accuracy and limitations: AI outputs are probabilistic and may contain errors or bias. SIKER monitors the performance of its AI systems and conducts periodic reviews as part of its development and maintenance processes.
• Fairness: SIKER incorporates periodic reviews to identify and mitigate potential biases or disparities in career recommendations and talent scoring.
• Human Oversight: SIKER maintains a human review function for AI results that have been escalated by users or that trigger automated risk alerts.

3. Human Supervision Requirements

SIKER requires or strongly recommends human supervision in the following contexts:

• Institutional Admissions Decisions: TalentOS scores must be reviewed by a qualified academic professional before being used as input in any admission, advancement, or scholarship decisions. SIKER contractually requires this through the applicable Data Processing Agreement with institutional clients.
• Employment and Recruitment Decisions: TalentOS talent profiles and competency scores should be reviewed by a qualified human resources professional or hiring manager. Automated rejection of candidates based solely on TalentOS results is prohibited under our Business Customer Terms.
• Content Publishing: Content generated by PathMaker must be reviewed by the user before being published. SIKER's terms require that users take responsibility for the content posted.
• Financial and Legal Decisions: No AI results from any SIKER product should be used as the sole basis for a financial, legal or medical decision.

4. Responsible Use Guidelines

Users and Business Clients agree to the following by using SIKER AI-powered Services:

• AI Outputs may not be presented as written works of independent human authorship in contexts where disclosure of AI participation is required by platform policies, professional standards, academic integrity codes, or applicable employment contracts.
• AI Outputs may not be used to unlawfully discriminate against any person or group on the basis of any protected characteristic.
• Business Clients may not use SIKER AI systems to evaluate candidates or students in a manner that violates applicable employment, education, or anti-discrimination laws.
• Users should not attempt to reverse engineer, extract model weights, or probe AI systems through systematic adversarial input.
• Users and Business Clients agree to report AI errors, biases or harmful results to ethics@siker.info so that SIKER can investigate and take corrective action.

5. AI Training and Data Usage

SIKER may use de-identified and pseudonymized data derived from user interactions to train and improve its AI models. The following rules apply:

• Personally identifiable data is not used for training AI models without your separate, explicit consent, which you may grant or revoke through your account Privacy Settings.
• Data from enterprise and institutional customers is not used to train models that benefit other customers without the express written consent of the Enterprise Customer.
• Third-party base models, such as GPT class models accessed via API, are used subject to the usage and privacy policies of the applicable provider. SIKER does not instruct these providers to train with customer data without explicit authorization.

6. Data Minimization and AI

SIKER applies data minimization principles to AI inputs. Our AI systems are designed to operate with the minimum personal data necessary to generate useful results. Where anonymized or aggregated data is sufficient to provide a useful result, we prefer that approach.

7. AI Governance

SIKER's AI governance framework includes:

• An internal review process involving engineering, legal, and product leadership prior to the release of high-impact AI capabilities.
• Internal documentation of each AI model in production, including its intended use, training data sources, and known limitations.
• A response process to address and investigate reports of possible algorithmic bias received from users or clients.
• Periodic internal audits of AI systems. As SIKER grows, it aims to incorporate independent reviews and share relevant findings with Enterprise Clients.

8. Limitations and Disclaimers

AI SYSTEMS ARE NOT INFALLIBLE. SIKER AI PRODUCTS DO NOT PROVIDE PSYCHOLOGICAL DIAGNOSES, CERTIFIED CAREER GUIDANCE, LEGAL ADVICE, MEDICAL RECOMMENDATIONS OR ACADEMIC CREDENTIALS. NO AI OUTPUTS FROM ANY SIKER PRODUCT SHOULD BE USED AS THE SOLE BASIS FOR ANY CONSEQUENTIAL DECISION AFFECTING THE EDUCATION, EMPLOYMENT, HEALTH OR LEGAL RIGHTS OF AN INDIVIDUAL. SIKER EXPRESSLY DISCLAIMS ANY WARRANTIES REGARDING THE ACCURACY, COMPLETENESS OR SUITABILITY OF AI OUTPUTS FOR ANY PARTICULAR PURPOSE.

9. Contact — Ethics and Transparency in AI

To report AI bias, request explainability information, or ask questions about our AI systems, contact: ethics@siker.info

ethics@siker.info