AI Governance & Ethics Policy

LST — LifescienceTalent.ai

Version: 1.0 Date: 2026-03-01 Review Cycle: Annually

1. Purpose

This document explains how LST uses Artificial Intelligence (AI) within the Life Science Talent Platform, what principles guide its use, and what rights candidates and clients have in relation to AI-based processing.

LST uses AI to help match life-science professionals with project opportunities — faster, more precisely, and more fairly than manual screening alone. We are committed to responsible and transparent AI use that always keeps human judgement at the centre of every significant decision.

2. EU AI Act — High-Risk Classification

Under Regulation (EU) 2024/1689 (EU AI Act), AI systems used in recruitment and employment decisions are classified as High-Risk AI Systems (Annex III, Point 4). LST takes this classification seriously and fulfils the corresponding obligations:

• Risk management — ongoing identification and mitigation of AI-related risks
• Data governance — controlled, documented use of personal data
• Transparency — clear information to candidates and clients about how AI is used
• Human oversight — every consequential decision is reviewed and made by a person
• Accuracy and robustness — regular monitoring and quality review of AI outputs
• Auditability — all AI decisions are logged and can be reviewed

The large language models (LLMs) we use are provided by established AI companies (including Google and Anthropic) who are subject to their own EU AI Act obligations as providers of General-Purpose AI models.

3. How We Use AI

LST uses AI at several stages of the recruitment process:

CV Analysis

When a candidate uploads their CV, AI automatically reads and structures its content — work history, skills, education, and more. This removes manual data entry and ensures that relevant profile information is available for matching.

Candidate–Job Matching

AI compares candidate profiles with available project positions based on meaning and context, not just keywords. The result is a ranked list of the best-fitting candidates for each position — and the best-fitting positions for each candidate.

Match Evaluation

For the most promising matches, an AI model produces a written assessment: what makes this candidate a strong fit, what questions should be explored, and a match quality rating.

For the best-fitting candidates, the matching position is presented to the candidate within the application process, and an AI screening agent is initiated to conduct a screening interview with the candidate — covering both administrative topics and any concerns. If the candidate confirms interest, the profile is presented to the client together with the screening answers and the AI's assessment. The client then independently decides whether to start the formal interview process.

Job Brief Assistance

Clients can use AI assistance when describing a new project role. The AI helps structure and complete the job brief, improving the quality of the search and the accuracy of subsequent matching.

Pre-Screening Conversation

For best-fitting candidates, the AI presents the matching position within the application process and initiates a screening agent that conducts a structured pre-screening interview with the candidate — addressing both administrative topics and any concerns. The AI assistant always identifies itself as an AI at the start of the conversation. If the candidate confirms interest, the profile is presented to the client together with the screening answers and the AI's assessment. The client then independently decides whether to start the formal interview process. A human recruiter reviews the screening results before any further steps are taken.

4. Human Oversight — Always

No candidate is accepted or rejected solely by AI. The AI produces recommendations; humans make decisions.

5. Data Protection & Your Rights

Where Your Data Is Stored

All personal data is stored on servers located in the EU (Frankfurt, Germany). When AI processing requires sending text to an external AI provider, this is done under a Data Processing Agreement that prohibits the provider from storing or using your data for their own purposes.

What Data Is Processed

AI processes the professional information in your CV:

• Work history (employers, roles, sectors, countries, dates)
• Education and certifications
• Skills and language proficiency
• Your stated preferences (availability, preferred work model, rate expectations)

Your name and contact details are stored separately and are not part of AI matching or evaluation.

We do not process special categories of personal data (such as health information, ethnic origin, or religious beliefs). Please do not include such information in your CV.

All Categories of Personal Data We Process

In addition to the professional information used for AI matching, LST processes the following personal data to operate the platform:

• Account & identity data — name, email address, password hash, account role (talent / client / admin)
• Profile & contact data — phone number, postal address, billing details where applicable
• Authentication & session data — login timestamps, session tokens, password-reset events
• Uploaded documents — CVs, supporting documents, contract drafts, signed contracts
• Communication data — messages exchanged with recruiters or screening assistants on the platform
• Technical & log data — IP address, browser/device user-agent, request logs, error logs, audit logs
• Usage data — pages visited, features used, application progress within the platform

This data is processed to provide the service, to fulfil contractual and legal obligations, and to secure and improve the platform.

Recipients & Processors

LST works with the following categories of recipients and processors. All processors act under a Data Processing Agreement (Art. 28 GDPR).

We do not sell personal data and do not share it with third parties for advertising purposes.

Cookies & Tracking

LST currently uses only strictly necessary cookies required to operate the platform:

• Session cookie — keeps you logged in for the duration of your session
• CSRF token — protects forms against cross-site request forgery
• Authentication cookie — remembers your login between visits if you choose "stay signed in"

No marketing cookies, no advertising trackers, and no third-party analytics are loaded on the public site. If we later introduce analytics or any non-essential cookies, they will only be activated after explicit consent in line with § 25 TTDSG / ePrivacy Directive, and you will be informed before any data is collected.

Third-Country Transfers

Some AI processing involves providers headquartered in the United States (Anthropic, Google). Transfers to these providers take place on the following legal bases:

• EU-US Data Privacy Framework (Adequacy Decision of 10 July 2023) — both providers are certified under the framework.
• Standard Contractual Clauses (Art. 46(2)(c) GDPR) — applied additionally as a fallback safeguard.
• Technical safeguards — input data is limited to what is necessary for the specific task; no special categories of personal data are transferred.

You should be aware that US authorities may, under laws such as FISA 702 and the CLOUD Act, in principle require US providers to disclose data. We minimise this risk by limiting the data sent, by contractual safeguards, and by storing the canonical copy of your data only in the EU. You can request that AI processing of your profile be limited; in that case, the matching quality may be reduced.

Legal Basis

Your data is processed on the basis of:

• Contract performance (Art. 6(1)(b) GDPR) — AI matching is a core part of the LST service
• Consent (Art. 6(1)(a) GDPR) — where processing goes beyond what is strictly necessary for the service
• Legal obligation (Art. 6(1)(c) GDPR) — audit logs required under the EU AI Act

Your Rights

You have the following rights regarding your personal data:

Data Retention

6. Why We Believe AI Makes Recruitment Fairer

We believe that AI, when used responsibly, can reduce some of the most common sources of unfairness in recruitment.

Human decision-making is affected by factors that have nothing to do with a candidate's professional fit: first impressions, the order in which CVs are reviewed, a recruiter's mood, unconscious associations with names, institutions, or career gaps. AI evaluates every profile against the same criteria, in the same way, every time — without fatigue, without distraction, and without the influence of how someone looks or sounds.

This does not mean AI is perfectly neutral. AI models learn from historical data, and historical hiring patterns reflect the biases of the past. A model trained on biased data can reproduce and even amplify those biases in subtle ways. This is a real risk we take seriously.

Our position: AI is a more consistent evaluator than humans for structured criteria — skills match, experience fit, availability, rate alignment. For these factors, AI is less susceptible to the day-to-day variability that affects human judgement. At the same time, AI lacks the contextual understanding, empathy, and situational awareness that an experienced recruiter brings to every placement decision.

This is why we use AI and humans together: AI for consistency and scale, humans for context and final judgement. Neither alone is sufficient.

7. Fairness & Non-Discrimination

LST is committed to fair and non-discriminatory use of AI.

Prohibited uses of AI on the LST platform:

• Filtering or ranking candidates based on gender, age, ethnic origin, disability, religion, sexual orientation, or nationality
• Automatically rejecting formally suitable candidates (whose CV matches the profile requirements) without human review
• Using AI assessments as the sole basis for any consequential decision

What we do to monitor fairness:

• Regular reviews of AI recommendation patterns to detect potential bias
• Review of AI model behaviour with every significant model update
• Recruiter feedback on AI quality is collected and used to improve the system

AI models — like all statistical systems — can reflect biases present in training data. We take this risk seriously and work to identify and correct it.

7. Transparency & Explainability

What You Are Told

As a candidate on the LST platform, you are informed:

• That your CV is analysed using AI
• That AI assists in matching your profile with positions
• That any pre-screening conversation may be conducted by an AI assistant, which will always identify itself
• That final decisions are made by human recruiters

What AI Assessments Contain

AI-generated assessments shown to recruiters include:

• A match quality rating
• A plain-language summary of the candidate's fit
• Specific strengths and potential concerns
• Questions to explore in a human interview

These are starting points for recruiter review — not final verdicts.

Limits of AI Explanations

AI matching is based on learned statistical patterns. While we provide qualitative explanations alongside every assessment, we cannot always provide a precise step-by-step account of why a specific score was generated. This is an inherent characteristic of modern AI systems; it is why human oversight is essential.

8. Auditability

Every AI assessment generated on the LST platform is logged with a full record of the input and output. This log:

• Enables LST to review any AI decision retrospectively
• Is available to regulators upon lawful request
• Can be summarised for candidates who request information about AI decisions affecting them

Logs are retained for 3 years in line with EU AI Act obligations.

9. Third-Party AI Providers

LST uses AI services from established providers. All providers are required to:

• Process your data only for the agreed service, not for their own purposes
• Not use your data to train their AI models
• Apply appropriate technical security measures
• Comply with GDPR as data processors (Art. 28 GDPR)

All primary data storage and processing remains within the EU.

10. Incident Response

If LST identifies a significant AI-related problem — such as a discriminatory output, a data breach, or a material system error — the following steps apply:

1. The affected AI component is immediately suspended
2. LST's data protection officer is notified within 24 hours
3. The supervisory authority is notified within 72 hours if personal data is involved (GDPR Art. 33)
4. The incident is investigated using audit logs
5. The system is corrected before being reactivated
6. The incident is documented in an internal register

11. Contact

For questions about AI-based processing, to exercise your data rights, or to report a concern:

• Email: admin@lifesciencetalent.ai

12. Version History

This policy is reviewed annually and updated when AI components or applicable regulations change significantly.