No Hacking Required: Why France’s E-Invoicing Mandate is a Gift to Foreign Rivals

April 2013, JFK airport. Frédéric Pierucci, an Alstom executive, is arrested by the FBI as he steps off his plane. The official grounds: bribery in Indonesia, prosecuted under an American law with extraterritorial reach, the FCPA. The practical outcome: two years in prison, a $772 million fine for Alstom, and the sale of the group’s energy business to General Electric, its American competitor, while its executive was bargaining for his freedom. Law as a weapon, a strategic French company as the trophy. Since then, no one can claim not to know that economic warfare exists and that France is among its targets.

Thirteen years later, that same France is requiring its own companies to pour the entirety of their invoicing flows into a centralized collection infrastructure, in the name of fighting VAT fraud. Millions of companies, their customers, their suppliers, their actual prices, their volumes, their rhythms. And at no point in the public debate has anyone asked the question that ought to open any national economic security council: what is this trove worth to a hostile intelligence service, and what have we done to protect it?

Economic warfare is not declared, it is waged

Some scene-setting first, because in French public debate the phrase “economic warfare” still passes for a figure of speech. It is an operational reality, complete with doctrines, budgets, and agencies. It is waged not with armored divisions but with law, data, standards, supply chains, financial sanctions, and now artificial intelligence.

The United States is its most unapologetic practitioner. The extraterritorial reach of American law delivered Alstom, but also an $8.9 billion fine imposed on BNP Paribas in 2014 for dollar-denominated transactions with countries under U.S. embargo. The Cloud Act subjects any data held by an American provider to U.S. judicial demands, wherever that data is hosted, a subject I have already covered in connection with Wero and its Amazon servers. And since the 1990s, American intelligence has openly embraced the mission of supporting national commercial interests: the Echelon affair, documented by the European Parliament, in which intercepted communications served the positions of Boeing and McDonnell Douglas against Airbus, remains the classic illustration.

China wages the same war with different weapons: massive investment in semiconductors and artificial intelligence, control of the rare earths whose use as an instrument of coercion I have described, and the methodical construction of interlocking dependencies in which every piece of infrastructure sold abroad doubles as a foothold. Europe, for its part, discovers the full extent of its dependencies with each new crisis: cloud, software, processors, platforms. I have already written about what the return of American protectionism will cost it.

In this permanent confrontation, economic information has changed in status. Knowing a rival’s supply chains, identifying its critical suppliers, anticipating trouble in an industry, or detecting an emerging competitor is no longer a byproduct of doing business: it is a resource of power, collected, aggregated, and analyzed as such.

What an invoice reveals to a hostile analyst

Now let’s take the exploiter’s point of view, because it is the only one that matters when assessing a risk. What does an economic intelligence analyst read in a stream of invoices?

The obvious things first: customers, suppliers, amounts. Then, very quickly, much better things. The prices actually charged, line by line, rather than list prices, which is to say margins and negotiating positions. Volumes and their seasonality, hence production rhythms and order books. The appearance of an unusual supplier, the weak signal of a prototype, an industrial pivot, an innovation in the works. Purchases concentrated with a single supplier, which is to say an exploitable dependency. Payment terms stretching out, an early symptom of tightening cash flow, information worth its weight in gold to anyone preparing a hostile takeover or a squeeze out of a market.

Taken in isolation, every invoice is banal. Aggregated and run through automated analysis tools, this data stops being accounting and becomes strategic knowledge. And the French system adds the dimension that changes everything: time. With continuous transmission of flows and mandatory e-reporting, what is being assembled is not a snapshot of the French economy but a near-real-time film of it. I showed in May that the reform was in effect building the complete graph of France’s private economy; I won’t retrace that demonstration here. What concerns me now is different: who might want that graph, and by what routes they could get it.

This information belongs to companies’ strategic assets, every bit as much as their patents or their manufacturing processes. The difference is that a patent is protected by a centuries-old legal arsenal, while the invoicing flow has just been made mandatory, centralized with third parties, and entrusted to an ecosystem whose resistance to a state-level adversary no one has audited.

From common crime to state intelligence

My May article dealt with the criminal threat: ransomware, data leaks, a multiplied attack surface. That threat has one paradoxical virtue: it is visible. Ransomware makes demands, paralyzes, announces itself. Economic intelligence does exactly the opposite: a successful exfiltration produces no symptoms, no press release, no complaint. The absence of scandal proves nothing; it is, if anything, the signature of a job well done.

And the French precedents are not fiction. In 2019, Airbus suffered a series of attacks carried out not head-on but by ricochet, through its subcontractors and service providers, with technical and certification documents as the apparent targets and suspicion falling on an intrusion group linked to China. In 2016, more than 22,000 pages of documentation on the Scorpène submarines built by DCNS, now Naval Group, ended up in the wild, compromising export positions for decades to come. Two cases, one lesson: you don’t attack the fortress, you attack its ecosystem, the peripheral link whose security is not commensurate with the value it handles.

That is exactly the architecture the reform has just generalized. Every French company is being assigned an additional intermediary in its chain of trust, an accredited platform through which its flows will transit. The Airbus playbook, attacking the subcontractors to reach the prime, now applies to the entire economy: the tier-2 and tier-3 suppliers of Naval Group, of Dassault, of the nuclear and space industries will invoice their services through these platforms, and their invoices will say what their contracts keep quiet.

The scenarios that don’t even require hacking

Hacking, in fact, is only the crudest scenario. A serious adversary has more elegant paths, and this is where the French system reveals its blind spot.

The legal lever, first. Let an accredited platform be the subsidiary of an American group, or host its processing with an American cloud provider, and the data it handles falls within the scope of the Cloud Act and of discovery proceedings. The accreditation file does require a commitment to operate the information system from within the European Union; the Cloud Act was written precisely to render that kind of precaution ineffective, since it attaches to the legal entity, not to the location of the servers. The precedent exists and it is massive: the TFTP program, through which the U.S. Treasury spent years exploiting the data of the SWIFT interbank network, a company incorporated under Belgian law. No implants, no exfiltration needed, when a subpoena will do.

The capital lever, next. Who owns the accredited platforms? Who will own them in five years, after the inevitable consolidation of a fragmented market? Buying an accredited operator means legally acquiring an observation post over the commercial relationships of thousands of French companies. Foreign investment screening applies to sensitive sectors; whether it actually covers this situation will have to be checked against the text, because platform accreditation rests on security and compliance requirements, not on any doctrine about the nationality of capital.

The human lever, too. A systems administrator at an invoicing operator sees more economically sensitive information pass through his hands than a ministry official, with no security clearance, no vetting, none of the legal framework that governs those who come near state secrets. Recruiting a source inside an underpaid private company remains the oldest and most cost-effective method in the trade.

The aggregate lever, finally. Even anonymized, even aggregated, fine-grained sector statistics reveal the health of an industry, its bottlenecks, its dependencies. And de-anonymization by cross-referencing sources has been a documented problem for twenty years: in an economy where many markets are local oligopolies, the aggregate becomes nominative again for anyone holding the right matching data.

The Italian objection, and why it should not reassure anyone

Italy will be raised against me, and the objection deserves better than a wave of the hand. The Sistema di Interscambio has centralized electronic invoicing there since 2019, with no known public catastrophe. Three answers.

The first is architectural: the SdI is a single public system, operated by the Italian tax administration. France has chosen a fragmented model, more than a hundred accredited private operators, which is to say that many heterogeneous security perimeters, hiring policies, ownership structures, and cascading layers of subcontracting. You do not compare a vault to an archipelago.

The second concerns the nature of the threat: as I said, the absence of scandal does not prove the absence of exploitation. The whole point of intelligence work is to leave no trace, and the Italian counterexample is at best a silence, not a proof.

The third concerns the target: France’s industrial base, defense, nuclear, space, aeronautics, dual-use, concentrates a patrimony that the Italian economy, respectable as it is, does not hold to the same degree. The graph of French subcontractors is, in itself, a piece of military intelligence.

There remains the internal objection, the most serious one, the one the reform’s architects will make: the state does not receive everything. Since the public portal was abandoned as an exchange platform in October 2024, the administration keeps only a central directory and a data concentrator, fed by a normalized extract of each invoice: some two dozen header fields and a handful of fields per line. The full document, with its precise line descriptions, its terms, and its attachments, never reaches the tax authority: it transits through, and resides with, the platforms. The objection is accurate, and it cuts the other way. It means that the richest layer of the trove, the one that interests a hostile analyst far more than the tax extract, is stored not in the administration’s fortress, under the legal regime of tax secrecy and the protection of a sovereign apparatus, but with a hundred-odd private operators of very unequal means. The tax administration already saw a great deal: VAT returns, the standard audit file, fee declarations. The current leap does not consist in showing it more; it consists in making the whole thing circulate, continuously, outside its walls. I have already analyzed what this system means from the taxpayer’s point of view; from a foreign power’s point of view, it means the data has left the fortress.

What a serious state would demand

The critique would be pointless if nothing could be done. And let’s be precise, because the accreditation requirements are not empty: mandatory ISO 27001 certification, SecNumCloud qualification (the French cybersecurity agency’s demanding cloud standard) required for subcontracted hosting, a commitment to operate the information system from within the European Union, a compliance audit after accreditation. But all of this protects systems, not contents, and it offers no defense against a shareholder, a foreign judge, or a human source. Encryption exists in this system only in transit, and it cannot be otherwise: the platforms’ very function, converting invoices between the Factur-X, UBL, and CII formats, extracting the tax data, validating the mandatory fields, requires them to read every document in the clear. End-to-end encryption is not missing by oversight; it is structurally ruled out by the chosen architecture.

A serious state would nonetheless have made it a founding requirement: envelopes encrypted from sender to receiver, which the platforms route without being able to open, with the normalized tax extract produced and sealed on the sender’s side. Nothing forbade it, neither the European standard EN 16931 nor the frameworks of ANSSI, France’s national cybersecurity agency; to my knowledge, no public initiative is even sketching such a model today. A serious state would likewise require that the capital of accredited platforms pass systematically through foreign investment screening, as it does for operators of vital importance. And it would finally recognize aggregated invoicing data as part of the nation’s strategic informational patrimony, subject to the same doctrine that protects the country’s scientific and technical potential, with an active role for the SISSE, the government service in charge of strategic information and economic security.

None of this sits at the core of the system. The asymmetry I pointed out in May remains intact: the state mandates everything, guarantees nothing, and will answer for nothing.

Sovereignty is not decreed in a requirements document

The most striking thing is the doctrinal contradiction. For fifteen years, France has been methodically building a defensive economic-warfare arsenal: reactivating the blocking statute of 1968, creating the SISSE, transposing trade secrets protection into law in 2018, continuously broadening its foreign investment screening. It paid to learn; Alstom was the lesson. And at the very moment this doctrine reaches maturity, France is digging, in the name of VAT collection, the tunnel under its own walls: a dynamic map of its economy, continuously updated, distributed among third parties whose capital and personnel fall under no protective doctrine whatsoever.

The question posed at the outset therefore has an answer, and it is an unpleasant one. By centralizing this data without protecting it in proportion to its value, we are not strengthening France’s economic sovereignty. We are assembling, at our own expense, under legal compulsion, and with the diligence of a well-disciplined people, the most complete economic intelligence trove in Europe. All that’s missing now are the visitors, and recent history teaches that they never wait to be invited.