The Fifth Eye · An Interlude

The Ghost in the Machine

One company. One algorithm. A new badge for every vertical it enters — immigration, narcotics, banks, and now the energy markets. What Giant Oak built, what GOST does, and the question ATHENA leaves behind.

ToreSays.com Investigative Series · Companion to Part V

Part Five left a contract on the table without opening it. A $5.2 million award — contract no. 89603022C0001 — from the Federal Energy Regulatory Commission to Giant Oak, for the completion of a system the company calls ATHENA. It sat in the piece as a contrast: the fully itemized engagement, set against the Bureau of the Fiscal Service line that has no published contract number at all. But a contrast is not an explanation, and the ATHENA contract raises a question Part Five deliberately set aside — one whose answer is older than the contract itself. The same founder who once helped DARPA turn battlefield data into targeting judgments now sells the civilian version of that engine to every vertical that needs to find the hidden actor in a crowd. Why is a company built to hunt human beings inside data now being paid to watch the energy markets?

One point of precision before we go further, because the name invites confusion. The ATHENA in this contract is not a corporate entity, a foreign holding group, or a consumer brand that happens to share the word. It is not the Dubai-based Athena Group of Companies, the crypto-and-e-commerce conglomerate run by the social-media personality Mario Nawfal — an unrelated operation that dominates a search engine’s results and has nothing whatever to do with this story. ATHENA, here, is strictly the name of a proprietary analytic system built by Giant Oak under a federal contract, developed and deployed inside the Washington defense corridor. It does not sell blenders or tokens. It watches people.

To answer that, you have to understand what Giant Oak actually is — not the marketing, not the procurement label, but the engine. Because the engine is the through-line of this entire series. It is the same thing wearing different uniforms, and once you can see it clearly, the customer list stops looking like a coincidence and starts looking like a pattern.

· · ·

The ghost in the name

Giant Oak's flagship product is called GOST — Giant Oak Search Technology. Say it out loud. The company knows exactly what it is saying. The pitch, delivered for years by founder and CEO Gary Shiffman, rests on a single observation: the open, searchable internet — the part a normal search engine indexes — is a thin skin over a much larger body of data. By Shiffman's own framing, conventional search reaches only a sliver of what is actually out there. Everything else is, in the most literal sense, invisible. Unindexed. Unsurfaced. A ghost.

What GOST sells is the power to make the ghost appear. To take a name, a face, a fragment of identity, and resolve it against the vast unindexed remainder — adverse media, watchlists, the deep web, the residue people leave without knowing they have left it — and return a verdict. Threat or not. Suspicious or clean. Worth a closer look, or not. Even in its own procurement language, GOST is not merely a search box: federal vendor descriptions cast it as a tool that builds a custom internet domain, re-indexes the open and deep web, and organizes the residue into triage — sorting entities by relevance and threat level.

It is a near-perfect emblem for the thing itself. Giant Oak: rooted, solid, lawful, visible above ground. GOST: the hidden made visible, the unseen pulled into the light and scored. A company named for the most grounded thing in the landscape, selling the ability to summon ghosts. Whether that pairing was deliberate or a programmer's joke, it captures the whole enterprise more honestly than any brochure: this is a business built on the gap between what you can see and what is actually there.

What the engine does

Strip away the verticals and GOST performs one act, repeated. It takes a population — visa applicants, account holders, traders, names on a list — and it finds the anomaly inside it. It resolves who is connected to whom through relationships no one declared. It scores behavior against a model of what "normal" looks like, and it surfaces whoever falls outside the curve.

The public-facing descriptions support three functions: search and triage, risk scoring, and entity prioritization. The analytic read of this series is that those three functions, taken together, amount to something with older and harder names. Entity resolution — figuring out that these scattered fragments are all the same person or firm. Network mapping — figuring out the undeclared links between them. Behavioral anomaly detection — figuring out which of them is acting strangely. Everything else is a question of which population you point it at.

One engine, many badges

This is the part that matters, and it is the reason this interlude exists. The work GOST does is indifferent to its target. "Find the anomalous actor in a population and map the hidden network behind them" is the same mathematical act whether the population is would-be immigrants, suspected narcotics traffickers, bank customers flagged for money laundering, or energy traders suspected of manipulating a market. The model does not care what the dots represent. It only cares that there are dots, and that some of them are connected in ways that were meant to stay quiet.

So the same core engine gets sold, vertical by vertical, into every corner of the federal government and the private sector that has a population to screen:

The badge it wearsThe population it watchesThe vertical
ICE · CBPVisa applicants, migrants, U.S.-citizen online speechImmigration / Border
DEASuspected narcotics networksCounter-narcotics
State DepartmentVisa and consular screening populationsDiplomatic security
U.S. Air ForcePersonnel / threat screeningMilitary
Banks (via Consilient)Account holders flagged for money launderingFinancial crime / AML
FERC (via ATHENA)Energy-market participants and tradersMarket integrity

Read that column on the right from top to bottom. Immigration, narcotics, diplomacy, military, banking, energy. Six unrelated domains of American life, policed by six unrelated bodies of law, with nothing in common except this: each is a population somebody wanted to watch, and each was handed to the same engine. The vendor did not need six inventions. It needed one transferable act — find the anomaly in the population, map the hidden relationship behind it — and six customers with a mandate, a database, and a budget.

The banking row is not a metaphor; it is the proof. The engine reached financial institutions two ways, both documented. First through Consilient, the venture Giant Oak founded with K2 Integrity in 2020 to port the same approach into anti-money-laundering — using federated machine learning, which trains one model across many banks' siloed data without the raw records ever leaving each bank. In published trials the approach cut false-positive alerts from over 90 percent toward roughly 12, while keeping the underlying data in place. Then, in February 2024, Fidelity Investments' compliance arm, Saifr, acquired GOST outright — the same platform that screens visa applicants for the government, repackaged as an adverse-media and sanctions-screening product for banks, with Giant Oak's own CEO describing the move as accelerating compliance solutions for financial institutions. Consilient is the engine wearing a banking badge. Saifr is the engine sold, by name, into the commercial market. The math travels. That is not inference; it is a press release.

That is the vendor estate this series keeps returning to. Not a conspiracy of aligned agencies, but a single private capability that has quietly become the connective tissue between agencies that otherwise share nothing — no mission, no statute, no oversight committee. The thing they share is a contractor. And the contractor's product is the surveillance of populations.

· · ·

Athena goes to market

Which brings us to the newest badge, and the one that should make you sit up: the energy markets.

First, the firewall — because precision here protects everything else. ATHENA is not GOST, and it is not a window onto the power grid. It does not, on any public evidence, watch how electricity is generated, how gas moves through a pipeline, or how power reaches a home. The FERC office that buys analytics like this, the Office of Enforcement, does not watch electrons. It watches people. Specifically, it watches the behavior of the firms and traders who buy and sell energy — hunting for spoofing, wash trades, cross-market manipulation, anomalous positions, and the concealed ownership relationships that hide who is really behind a trade.

FERC's own enforcement arm describes its Division of Analytics and Surveillance as building algorithmic screens over participant-level trading data to flag anomalies and identify investigative subjects — and it has openly complained that it sees too little of the financial affiliations standing behind those traders. The scale is industrial. In fiscal year 2025 alone, FERC's surveillance screens generated roughly 1,780 reviews in natural gas and 1,920 in electricity, narrowing to 24 and 36 deeper inquiries and a handful of referrals for investigation; staff combed nearly 2.7 million transactions from the quarterly reports filed by market sellers. The data environment behind those screens is larger still — non-public market data delivered by the organized markets under Order No. 760, and the eTags that document scheduled power interchange, accessible under Order No. 771. Read that against the GOST engine and the fit stops being a metaphor and starts looking like a hand in a glove. ATHENA may not wear the GOST label in public, but the functional shape is familiar: detect the anomaly, resolve the hidden relationship, surface the actor who was not supposed to be visible. The badge changed. The act did not.

It is the same act. Only the population changed.

It is worth being fair about this, because the legitimate reading is real. Market manipulation in energy is a genuine harm — it moves the price of the electricity and gas that ordinary people cannot choose not to buy. A tool that catches a trader rigging a winter price spike is doing something most people would applaud. FERC is supposed to police these markets, and policing them does require exactly this kind of analytics.

But that defense answers a smaller question than the one ATHENA actually raises. It explains why FERC requires anomaly detection to protect consumers from rigged prices; it does not explain why this specific contractor — whose lineage runs back through a DARPA war-zone targeting program and whose engine also screens migrants and bank customers — is the one pulling the lever in the wholesale gas and power markets, where finance, critical infrastructure, and national security sit closer together than almost anywhere else in American life. The claim is not that ATHENA is illegal or sinister. The claim is architectural: the same class of private machinery built for suspicion, scoring, and hidden-network discovery keeps being handed new populations under new legal labels, while the public is asked to evaluate each deployment as if it stood alone. Catching manipulators is the stated purpose. Whether it is the only purpose is what cannot be read off a procurement ledger.

Why energy

So: why energy? Why would an engine built to map illicit human networks gravitate, again and again, toward the energy markets? There are three honest answers, and they stack.

The technical answer. Because it is the same math. Energy-market manipulation is detected exactly the way illicit networks are detected — by resolving hidden relationships between players and flagging behavior that breaks the pattern. FERC needs to know who secretly controls and finances the entities trading in its markets. That is entity resolution and network mapping by another name. The engine did not have to be modified for energy. Energy was simply a population it could already read.

The commercial answer. Because energy is just the next vertical. To the vendor, there is nothing special about it — it is one more population with one more agency willing to pay to watch it. The business model is not "specialize in immigration" or "specialize in banking." The business model is "build the engine once, sell it everywhere there is a population and a budget." Energy was an open market the engine had not yet been sold into. Now it has.

The strategic answer — and this is where energy stops being ordinary. Because energy is not just another vertical. It is the one resource every other system depends on to function. Map the energy markets deeply enough — who controls supply, who finances the players, who moves which volumes at which moment of demand — and you are not merely policing a market. You are building a model of the circulatory system of the entire economy. The same visibility that catches a manipulator is, structurally, the visibility that lets you see leverage: who can be squeezed, who depends on whom, where the pressure points sit. A firm built to find the hidden actor in a network would gravitate to energy for the same reason water finds the lowest ground. Energy is not just a market; it is the dependency graph underneath the country. Every hospital, server farm, port, base, exchange, and home sits somewhere inside it. To map the markets around energy is not to own the grid — but it is to see the pressure points around the one thing every other system has to touch.

Here is the honest limit, and the series will hold to it. The pattern is documented: one analytic apparatus, colonizing vertical after vertical, arriving now at the most consequential market of all. What is not documented — what no contract will ever tell you — is intent. Whether anyone sat in a room and decided to unify these verticals into a single field of view, or whether a good product simply sold well in six directions, is the question the public record cannot answer. That is the seam between analysis and assertion, and it is exactly where reporting has to take over from the procurement database.

But the questions Part Five posed now have shape. If a system can already read every anomaly in how energy is traded, the distance to seeing how energy is distributed — by whom, to whom, at what moment of need — is a matter of scope, not of capability. The missing piece is not technical; it is jurisdictional. The same data pipes and scoring logic that flag an anomalous trader could, under a crisis statute or a new memorandum of understanding, be pointed at who sits behind a substation, a pipeline, or a retail provider. Whether that ever happens is not a thing the FERC contract decides. It is a thing worth watching for.

· · ·

The point of this interlude is not that ATHENA is sinister. It may be doing exactly and only what FERC says it does. The point is narrower and harder to unsee: there is one engine here, not six. It was built to surface the invisible — that is the whole meaning of the ghost in its name — and it has been carried, vertical by vertical, into immigration, narcotics, diplomacy, the military, the banks, and now the markets that power everything else. The Fifth Eye was never a foreign alliance. It is a vendor estate. And the estate just planted its flag on the most valuable ground on the map.

Sources & Documentation
  1. Federal contract award records, USAspending.gov / FPDS — Giant Oak, Inc. recipient profile and Federal Energy Regulatory Commission contract no. 89603022C0001 (“Completion of Development of ATHENA for FERC”), $5.2M, awarded Mar. 28, 2022. usaspending.gov
  2. Noah Shachtman, “Inside Darpa’s Secret Afghan Spy Machine,” Wired (Danger Room), July 21, 2011 — on DARPA’s Nexus 7 program. brookings.edu (reprint)
  3. Joseph Cox, “Inside ICE’s Database for Finding ‘Derogatory’ Online Speech,” 404 Media, Oct. 24, 2023 — FOIA-based reporting naming the federal agencies that have paid for Giant Oak / GOST. 404media.co
  4. “Consilient, K2 Integrity & Giant Oak Collaborate with Intel to Launch a New Approach to Fighting Financial Crime,” PR Newswire, Oct. 29, 2020; and “Consilient Bank Trials Demonstrate Federated Machine Learning…,” PR Newswire, Feb. 10, 2021 — on the federated-learning approach and trial results. prnewswire.com
  5. “Saifr Expands Tech Capabilities by Acquiring New Solution…,” Business Wire / Fidelity newsroom, Feb. 20, 2024 — Saifr (a Fidelity Investments company) acquires GOST from Giant Oak. newsroom.fidelity.com
  6. Federal Energy Regulatory Commission, FY2025 Report on Enforcement, Nov. 20, 2025 — Division of Analytics and Surveillance figures (surveillance reviews, inquiries, referrals; ~2.7M EQR transactions reviewed) and data-access authorities (Order Nos. 760 and 771). ferc.gov
ATHENA’s specific functionality is not described in the public contract record; its characterization here as a market-conduct surveillance build is an inference from FERC’s documented enforcement mission and Giant Oak’s documented analytic capability, and is labeled as such in the text.

Support Independent Journalism

Reader-funded. No advertisers. No editors. No leash.

Books by Tore

The Digital Dominion Series

Four volumes mapping the machinery they built around you.

Also by Tore

The Unedited History Project

The record they wouldn't keep, the history they wouldn't write.