Last updated: 03/03/2025
Published by Socially Recruited Ltd, trading as Gaia
At Gaia, our AI-powered talent attraction platform is designed to enable hiring teams to attract quality candidates at scale - ethically, transparently, and securely. This AI Compliance Statement outlines our commitments to responsible AI governance, bias mitigation, legal compliance, and enterprise-grade data protection.
1. Scope of AI Use at Gaia
GaiaAttract: Programmatic bidding optimisation, channel selection, dynamic audience targeting, retargeting logic
GaiaPages: Behavioural nudging, smart form sequencing, abandonment recovery
GaiaChat: Conversational AI for voice/chat-based candidate engagement, form data capture, and employment self-verification
GaiaTrack: AI-generated candidate profiles
Gaia does not make hiring decisions on behalf of customers. Final candidate selection always remains with the employer.
2. Core AI Ethics Principles
We operate under the following guiding principles, embedded in both product and process:
• Human Oversight: All AI-driven outputs are reviewable, reversible, and monitored by humans.
• Fairness & Inclusion: Our systems are trained and tested to reduce discriminatory outcomes.
• Transparency: Customers are informed of all AI use cases.
• Privacy by Design: Data protection is built into all AI workflows, in line with UK GDPR and ISO 27001.
• Purpose Limitation: AI models serve specific operational tasks and are tightly scoped to avoid generalisation risks.
3. Governance & Regulatory Compliance
Gaia’s AI Governance Programme is aligned with the following frameworks:
• UK ICO Guidance on AI and Data Protection
• Draft EU AI Act
• OECD Principles on AI
• NIST AI Risk Management Framework
• ISO 27001 (Certified)
• ISO 42001 (AI Management) readiness under review
An internal AI Risk Register and AI Governance Committee oversee responsible model use, approval processes, and impact assessments.
4. Model Training & Third-Party AI
• Gaia does not train any models on customer or candidate data.
• LLMs are only used for generative tasks and are not fine-tuned on proprietary or PII data.
• Gaia’s AI workers operate under strict prompt engineering and filtering protocols to ensure secure, non-intrusive, non-biased interactions.
5. AI Bias Mitigation Strategy
We apply proactive and reactive bias mitigation techniques including:
• Demographic-agnostic targeting for ads
• Performance-based audience optimisation over profile-based assumptions
• Behavioural trend mapping across networks instead of single-channel learning
• Content fairness checks on all AI-generated creative
• Quarterly bias audits across programmatic performance and ad delivery
Detailed bias approaches are available upon request.
6. Retargeting & Predictive Advertising
Gaia’s AI retargeting system operates without cookies or personal identifiers. It relies on behavioural overlaps across social platforms (e.g. engagement on TikTok influences targeting on Instagram), supported by:
• First-party and contextual signals
• 100M+ anonymised social graph data points
• Predictive analytics and auction-based performance forecasting
• Budget reallocation algorithms driven by ML thresholds
These techniques are used to attract interest, not profile individuals.
7. AI Risk Assessment & Impact Evaluation
• Gaia performs risk assessments on every high-impact AI feature.
• Where applicable, we support controller-led DPIAs with pre-filled responses.
• Models undergo testing for:
• Model drift
• Bias and performance degradation
• Accuracy of candidate interaction outputs
• Safety guardrails in generative systems (e.g. GaiaChat)
Risk levels are evaluated against regulatory thresholds (e.g. minimal, limited, high risk).
8. Explainability & Controls
Gaia enables:
• Review of bid strategy, ad switching, and performance changes driven by AI
• Full transparency on LLM-generated content (e.g. AI-written ads)
• Insight into why certain creative or channels were prioritised
• Candidate journey tracking through GaiaPages and GaiaChat, with conversion touchpoints explained
All AI systems are designed to be interpretable by non-technical users.
9. Security & Access Management
Gaia’s AI systems operate under our ISO 27001-certified ISMS and include:
• Role-based access control to all AI components
• Encryption of all data at rest and in transit
• Logging of all automated decisions and campaign actions
• Segregation of AI agents from core customer data pipelines
• Secure sandboxed environments for LLM interactions
10. Ongoing Compliance & Roadmap
• AI bias audits are performed quarterly
• Gaia’s AI Policy and model documentation are reviewed bi-annually
• Our Trust Centre will be updated with regulatory responses to the EU AI Act, UK AI Code of Conduct, and US Executive Order on AI
• A customer-facing AI Risk Summary and DPIA Support Pack is available upon request
Contact
For any questions about our AI usage or to request documentation, please contact:
compliance@iamgaia.com or visit our Trust Centre