The Death of the Data Autopsy: How Cryptographic Telemetry Can Ease the Sovereign Compliance Crisis

The autopsy problem

When a suspicious transaction is investigated in Nigeria today, the work usually starts long after the money has moved. Records are requested, exported, reformatted and reconciled across banks, payment service providers, microfinance banks and virtual asset service providers. By the time a picture forms, the funds have often crossed several institutions and, sometimes, a border.

We call this the data autopsy: careful analysis of something that is already over. It is not a failure of the people doing the work. It is a consequence of how records arrive: late, in different formats, through different channels, and without a shared way to show they have not been changed along the way.

Consider a simple illustrative case. Eleven transfers, each just under ₦1 million, leave a commercial bank account, a payment wallet and a microfinance account within fourteen minutes. They converge on one account at a virtual asset platform and are converted to stablecoin. Each institution sees a handful of ordinary looking transfers. Nobody sees the whole pattern until someone asks every institution for its records and stitches them together by hand. That stitching is the autopsy.

Pillar 1: A tightening regulatory landscape

Several developments, read together, push in the same direction.

Data protection. The Nigeria Data Protection Act 2023 sets duties around data minimisation, purpose limitation, security and cross-border transfer of personal data. Every extra copy of a BVN, NIN, phone number or account name is another copy an institution must justify and protect. Investigation workflows that move full customer files around by email or spreadsheet sit uncomfortably with those duties.

Data localisation and supervisory expectations. The Central Bank of Nigeria has long expected regulated institutions to keep and process customer and transaction data in ways it can examine, including expectations on where that data is hosted. In practice, supervisors still find it hard to confirm where records are written and replicated without on-site examination.

Identity exposure. Reports of identity records being resold cheaply, which we examined in The ₦250 Identity Leak, show that the risk often sits at integration points and downstream copies rather than inside central registries.

Virtual assets. The Investments and Securities Act 2025 brings digital asset activity more clearly into SEC supervision. P2P trades, OTC desks and cross-border stablecoin flows remain hard to see from any single institution, because the fiat leg and the crypto leg usually sit with different firms.

International standards. FATF Recommendation 29 describes a financial intelligence unit that receives, analyses and disseminates information. Recommendation 15 addresses new technologies and virtual asset service providers. Recommendation 16 addresses payment transparency, including information that travels with transfers.

The privacy debate abroad. In remarks at SIFMA on 23 September 2026, US SEC Commissioner Hester Peirce questioned whether collecting ever more customer data is the best way to find crime. Her remarks are a contribution to debate, not a rule, and not an endorsement of any vendor.

None of these require Onshorify. Together they raise a practical question: can supervisors get a faster, more trustworthy record without moving more raw identity data around?

Chronology at a glance

Readers should check each instrument directly for its current text. This table is a reading aid, not legal advice.

Pillar 2: From autopsy to signal

The approach we are building is simple to describe:

1. The institution keeps raw identity data inside its own perimeter.
2. Identity fields are replaced with a keyed token before anything leaves. The same customer produces the same token, so patterns can be linked, but the token cannot be reversed without the key and a lawful process.
3. A minimised transaction envelope, which we call stin.transport.v1, carries the token, amount band, timestamp, counterparty type and channel. It is signed by the institution.
4. The envelope also carries a signed write-origin attestation: the institution's statement of where the record was written.
5. Onshorify records when the envelope was received, checks the signature, compares the declared write origin with the observed network origin, and stores a SHA-256 digest so later alteration can be detected.
6. Deterministic rules flag patterns for supervisory review.

The rules are written down and testable, not a black box. Illustrative examples from our synthetic demonstrations:

• Fan-in velocity. Five or more credits from distinct institutions to one token within sixty minutes, each just below a reporting threshold.
• Zero dwell. Most of a balance leaving to an external rail within minutes of arriving.
• Coordinated devices. The same salted device fingerprint appearing across several institutions' apps in a short window.
• Origin mismatch. An envelope declared as written in Nigeria arriving from a non-Nigerian network origin.

A flag is a reason to look, not a finding. Thresholds would be set and reviewed with supervisors, not by us alone.

Pillar 3: Working with NFIU and goAML, not around them

The NFIU already runs reporting channels, including goAML, and institutions already carry statutory reporting duties. Telemetry should make those channels easier to feed, not compete with them.

• Structured inputs. Because envelopes are consistent, flagged sequences can be packaged in a shape that eases preparation of a report. Any mapping to goAML fields would be agreed and validated with the NFIU before use.
• Sovereign identity resolution. Onshorify works with tokens. Deciding who is behind a token remains with the NFIU and other competent authorities, through their lawful powers to request records from institutions.
• A clear boundary. Onshorify detects and packages patterns. It does not name individuals, adjudicate suspicion, or take enforcement action.

This matters for P2P and cross-border virtual asset activity in particular, where the useful signal is usually spread across a bank, a wallet provider and a VASP that do not see each other's data.

Pillar 4: The haystack problem and technical defensibility

Commissioner Peirce's point, in plain terms, is that a bigger haystack does not automatically make the needle easier to find, and it creates new risks for everyone whose data is in it. A central store of plaintext identity is an attractive target and a heavy compliance burden.

Telemetry takes a different route. The supervisory record holds tokens and digests, not names. If it were breached, the attacker would hold pseudonymous patterns, not a customer list.

On evidence, Section 84 of the Evidence Act 2011 sets conditions for admitting computer-generated records, including that the system was operating properly and the record was produced in the ordinary course. A signed envelope, an independent receipt time and a stored digest can help an institution or supervisor show those conditions. They do not decide admissibility. That is for the courts.

Pillar 5: The triple win

• Institutions carry fewer copies of sensitive data outside their walls and less manual work when a request arrives.
• Founders and compliance teams can show supervisors a clear, minimised data path early, rather than retrofitting one under pressure later.
• Supervisors get a timelier, tamper-evident record while keeping full control of identity resolution and enforcement.

None of these gains have been measured in production yet. They are the outcomes a pilot should be designed to test, against a baseline agreed in advance.

Old regime and telemetry era, side by side

What this does not do

Honest limits matter more than bold claims, especially before a regulator:

• A signature proves who sent a record, not that it is true. We reduce this risk by matching both legs of a transfer across institutions, by keeping signing keys in hardware, and through supervised inspection of the signing module against an approved build. No design removes it entirely.
• Origin checks show where telemetry was written from, not where every copy of the data is stored. Storage location still needs examination of contracts, infrastructure, replication logs and egress controls.
• Evidential weight is for the courts. A tamper-evident record can support Section 84 conditions; it does not decide admissibility.
• goAML and existing reporting duties remain. Telemetry does not replace an institution's obligation to file reports or the NFIU's choice of channels.
• There is no live integration with any agency today. Everything shown in Onshorify demonstrations runs on synthetic data.

Where to start

Teams that want to see where they stand can take the free Onshorify diagnostic. It takes a few minutes, works for banks, VASPs, startups, enterprises and regulators, and returns a plain-language readout of current gaps. The autopsy will not disappear overnight, but it does not have to remain the default.

All examples and figures in Onshorify demonstrations use synthetic data. References to statutes and public remarks are for context and do not imply endorsement by any authority.