What is said in the booth
In a bilingual negotiation, the interpreter is the only person in the room who understands everything. They are also the only one whose identity nobody checks.
Take a summit. A plenary hall, delegations, headsets. In the back, two glass booths where three people take turns. They hear the negotiating position before it is stated publicly, the figures that do not appear in the communiqué, a minister's hesitation, the phrase an adviser whispers aside.
Now ask the question that is almost never asked: who are these three people, who hired them, and how do we know?
At the vast majority of international events organised on the continent, the answer fits in one sentence: an organiser posted in a WhatsApp group, someone recommended someone, and the deal was done in a few messages.
An asymmetry the sector does not look at
Compare the treatment given to different service providers at the same summit.
The caterer undergoes a background check and wears a badge. The security company provides a nominative list of its agents. The IT provider signs a brief, sometimes undergoes an audit, and its network access is logged. The printer producing confidential documents works under a contract with a destruction clause.
The interpreter, by contrast, receives an email with the speeches as attachments.
This asymmetry has no rational justification. It comes from the fact that the sector classifies interpreting among logistics services — on a par with sound engineering — rather than in the chain of sensitive information processing, where it entirely belongs.
Map of the real risk
Five points of leakage, in the chronological order of an event.
Recruitment. The interpreter's identity, qualifications, experience and potential conflicts of interest are almost never verified. An interpreter working regularly for one party to a negotiation can end up in the booth for the other party, with no one being the wiser. This is not a textbook hypothesis; it is the mechanical consequence of a market where sourcing happens by word of mouth in a small pool.
Preparation documents. This is the most serious point, and the most commonplace. To interpret correctly, one must prepare: speeches, agendas, draft agreements, financial tables, technical notes. These documents often circulate several days before the event, by email or consumer messaging, to personal accounts, without encryption, without expiration, without logging. They then remain on personal computers for years. An attacker targeting a summit has no reason to attack the organiser's network: targeting the interpreter's email inbox is enough.
The booth and the audio stream. Remote interpreting has become the norm since 2020. A growing share of assignments now goes through platforms hosted outside the continent, chosen for their price, whose jurisdiction, retention policy and handling of recordings the organiser does not know.
Recording. Who records, on what medium, for how long, with what access control? In most cases, nobody can answer.
After the event. The confidentiality agreement, when one exists, is signed on paper, between two entities sometimes located in different jurisdictions, with no mechanism of proof and no practicable recourse. It produces legal comfort, not security.
It is not a deontology problem
We must be precise, because the nuance changes everything.
Professional conference interpreters are, in their overwhelming majority, rigorous. The profession very early on developed demanding ethics, and professional confidentiality is a cardinal value, genuinely respected. The problem is not the morality of practitioners.
The problem is that a system founded solely on individual ethics, without verification, without traceability and without technical safeguards, offers no verifiable guarantee. It works as long as everyone is honest and competent. It does not withstand error, pressure, or a malicious actor presenting themselves as an interpreter.
In security, you do not evaluate intentions. You evaluate attack surfaces.
What correction would cost
Very little, and that is what makes the situation absurd.
Verifying an interpreter's identity and references is a proven administrative process. Distributing preparation documents via an encrypted space, with named access, automatic expiration and access logging, is a banal technology available for fifteen years. Recording who accessed what and when requires no innovation. Choosing a remote interpreting platform whose jurisdiction and retention policy are explicit requires only a clause in the brief.
None of these elements belongs to research. They all belong to organisation. If they are not implemented, it is because nobody in the event's decision chain has responsibility for linguistic security. It is not the job of the organiser, the AV provider, or the translation agency.
Why this particularly concerns Africa
The continent hosts a growing number of summits, investment forums and sectoral negotiations — mining, energy, infrastructure, regional security. These are contexts where information has immediate, identifiable commercial value.
They are also, by construction, massively multilingual contexts. An African Union meeting mobilises several working languages. A mining negotiation in Central Africa may involve French, English, Mandarin and a national language. The longer the language chain, the more links it has, and the heavier each unverified link weighs.
In other words: the region of the world where dependence on interpreting is greatest is also where its security is least structured.
A question to ask
To summit organisers, institutional secretariats and legal directors who approve these services, I propose a single question, to ask at the next event: can you tell me, by name, who had access to the preparation documents, on what date, and through which channel?
If the answer does not exist, the meeting's confidentiality is not assured. It is hoped for.