Jobtrain and responsible AI in recruitment

Executive summary 

Recruitment is one of the most sensitive applications of artificial intelligence (AI). In the UK and EU, recruitment AI is classed as “high-risk” under the EU AI Act, and UK regulators (ICO, EHRC, CMA) are embedding fairness, transparency, accountability and human oversight into AI governance. 

Jobtrain provides an applicant tracking system (ATS) and onboarding platform trusted across a wide range of sectors. Our guiding principle is to deliver pioneering recruitment technology that keeps people at the heart of the recruitment process. 

We take a safety-first and compliance-led approach to AI. All our AI-powered tools are optional (clients can opt to disable them), assistive, bias-reducing and fully human-led, and none of them are trained on your data. 

This document sets out our AI features, our ethical principles, our compliance measures and our roadmap for responsible AI. It is intended for our clients, their IT colleagues and HR leaders, candidates and regulators seeking assurance of Jobtrain’s safe and ethical AI practices. 

Our AI principles 

Fairness and inclusion 

• AI features are designed to reduce bias (for example, job advert bias checking and structured assessment question generation). 

• We do not use AI to screen out or exclude candidates based on protected characteristics. 

Transparency and explainability 

• All AI features are optional, transparent and clearly explainable. Outputs are always editable by humans. 

• Recruiters understand how each suggestion is generated and retain full oversight of it. 

Human oversight 

• AI in Jobtrain supports recruiters but never makes final selection decisions. 

• Human review is required for shortlisting, assessments and appointments. As AI features evolve to support recruiters with more of the process, human review will also be required for any action an AI feature proposes on a user’s behalf. 

Privacy and data protection 

• GDPR-compliant data handling throughout. 

• Candidates retain control over their personal data. 

No model training, ever 

• Client and candidate data, including CVs, application content, assessment responses and messages, is never used to train, fine-tune or otherwise improve any AI model. This applies equally to models Jobtrain may develop and to any third-party AI model used within our platform. 

• Where a Jobtrain feature calls a third-party AI provider to generate an output, that provider is contractually restricted to processing the data solely to return the requested result, with no retention for model training or any other purpose. 

Accountability and compliance 

• Clear contractual agreements define responsibilities between Jobtrain and clients. 

• Regular Data Protection Impact Assessments (DPIAs) are conducted, including before any new AI feature is released. 

What Jobtrain does not do 

To protect candidates and clients, Jobtrain makes the following clear commitments: 

• No automated hiring, rejection or shortlisting decisions using AI: a person always makes the final call. 

• No profiling of candidates based on protected characteristics. 

• No use of client or candidate data to train, fine-tune or improve any AI model (ours or any third party’s) under any circumstances. 

• No indefinite data retention: all data is handled under GDPR’s storage limitation principle. 

• No AI feature or AI assistant connector is given the ability to take an action in Jobtrain, such as raising a job, progressing a candidate or sending a communication, without a person approving that specific action first; every such action is logged in a full audit trail. 

 

Governance and compliance 

• Data Protection Impact Assessments (DPIAs): conducted on all AI features before release, and refreshed as features change. 

• Audit trails: every recruiter action, and every AI-assisted suggestion or action, is logged for transparency. 

• Third-party AI providers: any AI or large language model provider used within Jobtrain is contractually required not to use client or candidate data to train its models, and processes data only to return the specific output requested. 

• Self-service candidate rights: candidates can withdraw, delete or track their data at any time. 

• Contractual clarity: Jobtrain’s and clients’ roles in data processing are clearly defined. 

• Independent review: our approach is reviewed for alignment with ICO guidance and EU AI Act obligations as they develop. 

Security and data protection 

• GDPR-compliant architecture throughout the platform. 

• Data minimisation: only necessary data is collected. 

• Two-stage application process allows sensitive data to be collected later if required. 

• CV parsing is optional, with candidates in full editorial control before submission. 

• Secure hosting and encryption, with no unauthorised third-party access. 

• AI assistant connectors are read-only by design, scoped to what each client authorises, and can be revoked by the client at any time. 

Transparency for candidates 

• Clear privacy and data use statements are embedded within the candidate journey. 

• Full audit trails ensure candidates can request feedback or challenge outcomes. 

• Candidates can exercise their GDPR rights at any time (access, correction, deletion, withdrawal). 

 

Alignment with regulations and standards 

• UK GDPR & Data Protection Act 2018 – Compliance is embedded in all workflows. 

• ICO guidance on AI and data protection – Transparency, fairness and explainability are built into every AI feature. 

• EU AI Act – High-risk obligations for recruitment AI were due to apply from 2 August 2026; the EU has since agreed to delay this to 2 December 2027. Jobtrain is proactively aligning ahead of that date rather than waiting for it. 

Future roadmap 

Jobtrain will continue to: 

• Monitor changes to UK and EU AI regulation, including the evolving EU AI Act timetable. 

• Publish updates for clients and candidates on any new AI features and any changes in legislation. 

• Introduce further bias-reducing and time-saving AI tools, each with full human oversight and each preceded by a DPIA. 

• Hold the line as AI features evolve towards assisting with more of the process, including features that can propose actions such as raising a job or progressing a candidate: every consequential action will still require a person to approve it, and every action will still be logged. 

 

Conclusion 

Jobtrain provides safe, effective and compliant AI tools that save time, reduce bias and support fairer recruitment. Our platform is assistive and human-led, never replacing professional judgement, and it is never trained on your data. 

By embedding fairness, transparency, accountability and compliance into our AI design, Jobtrain gives clients and candidates confidence that recruitment technology can be both innovative and ethical. 

Annex: mapping Jobtrain to regulatory standards 

Requirement	Jobtrain approach
GDPR Art. 5 – Lawfulness, fairness, transparency	Privacy-first design; candidate data rights; clear notices at every stage of the candidate journey.
GDPR Art. 22 – Automated decision-making	No automated decision-making. A human always makes the final recruitment decision.
ICO AI guidance – Explainability	AI features are documented and explainable; recruiters retain full control and can see how a suggestion was generated.
UK GDPR / Data Protection Act 2018 – Purpose limitation	Client and candidate data is never used to train, fine-tune or improve any AI model, Jobtrain’s own or a third party’s.
EU AI Act – High-risk AI systems (recruitment)	Obligations for recruitment AI, due to apply from 2 August 2026, have been delayed by EU agreement to 2 December 2027. Jobtrain is proactively aligning ahead of that date: human oversight, DPIAs, audit trails and bias monitoring are already in place.