EU AI Act Article 50 · Compliant
AI with a human signature
Every ThriveFinity output discloses its AI involvement, names its human verifier, and complies with EU AI Act Article 50. Here is exactly what that means — and what your rights are.
What the law requires — and how we comply
Article 50 of Regulation (EU) 2024/1689 — the EU AI Act — requires anyone deploying an AI system to generate content for human consumption to clearly disclose AI involvement. That disclosure must be machine-readable and perceptible to the recipient.
ThriveFinity has complied with the spirit of this obligation since launch. Every deliverable includes a standardised Article 50 Disclosure Block in both human-readable prose and structured schema.org metadata.
This applies to every service tier without exception. There are no opt-outs — disclosure is mandatory on every output we sign, regardless of the client or use case.
Non-compliance is not a risk we carry. Every output in our pipeline goes through a verification gate before delivery. If the disclosure block is missing, the output does not leave our systems.
Three stages. Every output signed.
Every stage is logged. No exceptions.
Retrieval
AI searches licensed data corpora, regulatory databases, and structured knowledge bases — not the open web. Every source is logged with a retrieval timestamp. No unverified web scraping.
Human verification
A named human verifier reads every claim, checks it against the cited source, rejects unsupported assertions, and rewrites where necessary. AI drafts; humans decide.
Signed disclosure
Every output includes the Article 50 Disclosure Block: AI model name, retrieval date range, and the human verifier's name and signature. Machine-readable and human-readable.
What AI does not do at ThriveFinity
These are structural constraints — not policies. Enforced by process design, not individual discretion.
Make investment, financial, legal, or medical recommendations
Produce regulated advice of any kind
Generate anonymous outputs — every output is human-signed
Use scraped, unlicensed, or unattributed data sources
Train on your confidential documents or briefs
Publish any output without a named human verifier review
Substitute for a qualified lawyer, accountant, or financial advisor
Deliver a verdict with unverified or miscited source claims
“Every ThriveFinity output is signed by a named human. If you receive an output without a human signature, it is not a ThriveFinity output.”
ThriveFinity Accountability Standard · Est. 2025
How we handle your data
We hold only what we need. We keep it only as long as necessary. We never sell it, share it, or train on it.
No training use
Your brief, data, and documents are never fed into any AI model for training — not ours, not our providers'. Client data is structurally isolated from all model pipelines.
Retention period
Project data is retained for 90 days post-delivery to support follow-up queries. After 90 days, documents and briefs are permanently deleted from our systems. Billing records are retained 7 years per UK tax law.
Request your data
You have the right to access, correct, or delete all data we hold about you at any time. Email privacy@thrivefinity.uk — we respond within 72 hours.
Project delivered. Data held securely.
Project data permanently deleted.
Billing records retained (UK law only).
You are in full control
GDPR and the UK Data Protection Act 2018 give you concrete rights over your personal data. Here's how to exercise them with us.
Access & portability
Request a copy of all personal data we hold. We'll send it in a machine-readable format (JSON or CSV) within 30 days. Email privacy@thrivefinity.uk with subject line "Data Access Request".
Correction & erasure
Ask us to correct inaccurate data or erase it entirely ("right to be forgotten"). We'll confirm deletion within 72 hours and send you a deletion certificate. Billing records cannot be erased earlier than 7 years under UK law.
Objection & complaints
You can object to any specific processing of your data. If you believe we've mishandled your data, you have the right to lodge a complaint with the Information Commissioner's Office (ICO) at any time.
Questions about how our AI works
Which AI models does ThriveFinity use?
We currently use Anthropic Claude (Sonnet / Opus tiers) for analysis and drafting, and GPT-4o for cross-validation passes. The specific model version and run date are included in every output's Article 50 Disclosure Block.
Can AI make the final call on a verdict?
No. Every verdict — GO, KILL, PIVOT, CONDITIONAL GO, or DEFER — is reviewed, edited, and signed by a named human verifier before delivery. AI surfaces evidence; humans decide.
Is my brief used to train any AI model?
Never. Your data is structurally isolated from all model training pipelines — ours and our providers'. This is a contractual commitment, not just a policy. See our Data Handling section above.
What is the Article 50 Disclosure Block?
A standardised disclosure appended to every ThriveFinity output. It states: (1) which AI models were used, (2) the date range of retrieval, (3) the human verifier's name, and (4) a brief description of the human oversight applied. Required by EU AI Act Article 50.
How do I report a concern about AI use?
Email ai@thrivefinity.uk. We treat every concern as a Priority 1 issue — you'll receive an acknowledgement within 4 hours and a substantive response within 48 hours.
Human intelligence. Machine speed.
Every ThriveFinity output is verifiable, signed, and accountable. See how that changes what's possible for your next strategic decision.
Questions? ai@thrivefinity.uk