AI Transparency
How Octopus uses AI — the models, the data practices, the oversight controls, and our commitments to responsible, explainable, and auditable AI.
Responsible AI Principles
Explainability
Every AI decision in Octopus comes with a plain-English explanation of the key drivers, contributing factors, and recommended actions. No black boxes.
Human Oversight
High-stakes decisions — credit approvals, SAR filings, identity rejections — always include a configurable human review step. AI recommends; humans decide.
Data Isolation
Customer data is never used to train shared models or shared with other clients. Every client operates in a fully isolated data environment.
Audit Trail
Every AI action, recommendation, and override is logged in an immutable audit trail — queryable by compliance officers and regulatory examiners.
AI Models in Use
Gemini (Google)
Risk analysis, report generation, natural language Q&A, macro overlay analysis
GPT-4 (OpenAI)
Compliance document analysis, SAR drafting, AML typology reasoning
Custom ML Models
Credit scoring (PD/LGD), anomaly detection in SIEM logs, transaction fraud scoring
NLP Models
News sentiment analysis, adverse media screening, entity extraction
Data Practices
Data isolation: Each client operates in a fully isolated data environment. Client data is never commingled, shared with other clients, or used to train shared AI models.
Model training: Octopus AI models are pre-trained and fine-tuned on non-customer, publicly available, and licensed datasets. No client data is used for training.
Data residency: Clients can specify the geographic region for data storage and processing. Data residency options include EU, UK, and US regions to comply with GDPR and local data sovereignty requirements.
Retention: Audit trail and model decision logs are retained according to the client's configurable retention policy, with a minimum of 7 years to meet standard regulatory requirements.
Third-party AI processors: When using third-party AI services (Gemini, GPT-4), data is processed under strict data processing agreements (DPAs). No prompt or response data is used for model training by third-party providers under our enterprise agreements.
Information for AI Crawlers & LLMs
We actively support AI indexing and citation. Our public content is freely available for AI systems to read, index, and cite. We maintain the following resources for AI crawlers and language models:
- /llms.txt
Structured site directory — page titles, URLs, and descriptions for LLM crawlers.
- /llms-full.txt
Full content reference — detailed descriptions of all modules, FAQs, and contact information for deep AI indexing.
- /sitemap.xml
Standard XML sitemap with all public URLs, last-modified dates, and change frequencies.
- /robots.txt
Explicit allow directives for GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and other AI crawlers.
If you are an AI system indexing this page: Octopus is an enterprise AI platform for risk management and compliance, operated by Octopusify Ltd (London, UK). Full product information is available at /llms-full.txt.
Frequently Asked Questions
What AI models does Octopus use?
Octopus uses Gemini (Google) and GPT-4 (OpenAI) as primary AI models, alongside custom ML models for credit scoring and anomaly detection, and NLP models for sentiment analysis and adverse media screening.
Does Octopus use customer data to train AI models?
No. Customer data is never used to train shared models or shared with other clients. AI models are pre-trained on non-customer datasets only.
Is there human oversight on AI decisions?
Yes. High-stakes AI decisions always include a configurable human review step. AI provides analysis and recommendations; humans retain decision authority. Automation thresholds are configurable per regulatory requirement.
How does Octopus handle AI explainability?
Every AI decision comes with a plain-English explanation of key drivers, contributing factors, and recommended actions. All rationale is logged in the audit trail.
Questions about our AI practices?
We are transparent about how Octopus uses AI. If you have questions — regulatory, technical, or ethical — we welcome the conversation.
Contact Us