Baritex AI Operating System AI Usage & Safety Policy

1. Responsible AI principles

• Lawfulness, fairness and respect for human rights.
• Meaningful transparency about AI involvement and limitations.
• Human responsibility for consequential decisions.
• Privacy, security, accessibility and data minimisation.
• Proportionate testing, monitoring, reporting and remediation.

2. What AI features do

Baritex AI Operating System may analyse prompts and files, generate drafts, diagnostics, recommendations, reports, career materials, business materials or workflow suggestions. Features and models may change. Output is probabilistic and may be inaccurate, incomplete, biased or unsuitable.

3. User responsibilities

• Review and verify important output before use.
• Do not submit data you lack authority to process.
• Use qualified professional advice for legal, medical, tax, financial, employment or regulated decisions.
• Disclose AI assistance where law, contract, professional rule or context requires it.
• Report harmful or materially incorrect output through support.

4. High-impact uses

AI output must not be the sole determinant of employment, credit, insurance, healthcare, legal rights, access to essential services or another similarly significant outcome where human review or safeguards are required. Users must evaluate relevance, accuracy, bias, accessibility and lawful basis.

5. Disability, accessibility and fairness

• Do not infer disability or health status unless lawful, necessary and appropriately safeguarded.
• Do not disadvantage a person because of disability, use of assistive technology, communication style or requested accommodation.
• Provide an accessible route to challenge material AI output and request human review.
• Baritex will progressively test key journeys and generated outputs for accessibility and material discriminatory effects.

6. Prohibited AI uses

AI must not be used for unlawful surveillance, exploitation, malicious code, fraud, phishing, harmful impersonation, deceptive deepfakes, targeted harassment, child sexual exploitation, unlawful discrimination or circumvention of safety controls. The Acceptable Use Policy provides further restrictions.

7. Data use and model training

Customer Content may be processed by authorised providers to generate requested output. It is not used to train general-purpose or shared models unless the customer affirmatively opts in or a separate written agreement expressly permits it. Baritex will apply contractual, privacy and security safeguards appropriate to provider risk.

8. Human review and reporting

A user may report unsafe, discriminatory or materially inaccurate output and may request human review where an outcome could significantly affect them. Baritex aims to acknowledge complaints within one business day and provide a substantive response or progress update within three business days.

9. Safety controls and enforcement

Baritex may apply content filters, usage limits, monitoring signals and human investigation to protect users and systems. Controls can make mistakes. Baritex will consider context, severity and accessibility needs and provides a review route before or after proportionate enforcement as circumstances permit.

10. Changes and accountability

Baritex will review this Policy as products, providers and law evolve. Material changes will be communicated appropriately. Legal & Compliance owns policy coordination; product and engineering owners are responsible for implementing applicable controls.