Responsible AI Policy

Last updated: 28 July 2026

At Baseflare Solutions Ltd (“Baseflare”, “we”, “us”), we are committed to the responsible development and deployment of artificial intelligence. This policy outlines how we use AI across our products and services, and our commitments to transparency, fairness, and human oversight.

How We Use AI

Baseflare develops AI-powered products for regulated industries. Our AI capabilities include:

  • Document classification identifying document types from uploaded files using machine learning models
  • Text extraction reading and structuring data from PDFs and images using optical character recognition
  • Validation and cross-referencing comparing extracted data against defined criteria and identifying discrepancies
  • Intelligent assessment using large language models to evaluate completeness and consistency of submissions

Our AI systems are built on AWS infrastructure using services including Amazon Textract, Amazon Bedrock, and custom machine learning pipelines.

Human Oversight and Decision-Making

Our AI products are designed as assistive tools, not autonomous decision-makers. We believe that consequential decisions affecting individuals must remain with qualified professionals.

  • AI outputs are presented as suggestions and flags for human review, never as final decisions
  • Users retain full responsibility for all decisions made using our products
  • All AI-generated assessments can be overridden by the user at any stage
  • We clearly distinguish between AI-generated content and verified factual data

Accuracy and Limitations

AI and machine learning models generate outputs based on patterns in data. While we strive for high accuracy:

  • AI outputs are probabilistic and may occasionally produce incorrect results
  • Input quality (document scans, image resolution, handwriting) affects output accuracy
  • Users should always verify AI-generated results before acting on them
  • We continuously monitor and improve our systems to maximise accuracy

Data Privacy and Security

We take the privacy and security of data seriously:

  • No data retention for model training documents processed through our products are not used to train or improve third-party AI models
  • Tenant isolation each customer’s data is logically and physically separated
  • UK data residency all data is processed and stored within AWS EU (London) region
  • Encryption data is encrypted at rest and in transit using industry-standard protocols
  • Access controls strict role-based access ensures only authorised personnel can access customer data

Prohibited Uses

Our AI products must not be used:

  • To make automated decisions about individuals without qualified human review
  • To discriminate against individuals based on protected characteristics
  • To fabricate, alter, or misrepresent content or verification results
  • To process personal data without appropriate consent or legal basis
  • To circumvent regulatory requirements or compliance obligations
  • For surveillance, profiling, or tracking of individuals without consent

Fairness and Bias Mitigation

We are committed to ensuring our AI systems treat all users and data subjects fairly:

  • Our systems are designed to be objective and consistent regardless of demographics
  • We regularly review AI outputs for potential bias
  • Where bias is identified, we take prompt corrective action
  • We select and evaluate AI models with fairness as a key criterion

Regulatory Compliance

Our products are designed to operate in compliance with applicable UK and EU regulations, including:

  • UK General Data Protection Regulation (UK GDPR)
  • Data Protection Act 2018
  • EU Artificial Intelligence Act (where applicable)
  • FCA guidance on the use of AI in financial services (where applicable)

Users are responsible for ensuring their own use of our products complies with all applicable laws, rules, and regulations.

Transparency and Accountability

We believe in being open about how our AI works:

  • We clearly indicate where AI is being used within our products
  • Results include confidence indicators where applicable
  • We maintain audit trails of AI-assisted decisions
  • This policy is reviewed and updated regularly as our technology and the regulatory landscape evolve

Contact

If you have questions about our Responsible AI practices, please contact us at hello@baseflare.co.uk.