A Protocol With No Deployment Clause
On August 31, 2026, Egypt's National Telecommunication Institute (NTI) and the State Lawsuits Authority — the body that represents the Egyptian state in litigation and drafts its contracts — signed a cooperation protocol expanding their joint work on digital transformation, artificial intelligence, cybersecurity, and digital capacity building for judicial personnel. NTI director Dr. Ahmed Khattab and State Lawsuits Authority vice-president and head of its Centre for Judicial Studies, Counselor Ahmed Saad Abdel Ati, signed it in Cairo, witnessed by Communications Minister Raafat Hendi and Authority president Counselor Abdel Nasser Abu El-Azm Eissa.
The headline mechanism is the "AI Ambassadors" programme, which NTI first ran for the Authority in 2025: more than 600 counsellors from across Egypt's judicial bodies completed that inaugural cohort, and a second, expanded cohort has now launched under the new protocol (Middle East Observer). NTI separately runs a general-public version of the same Ambassadors format — a 36-hour, 5th-cohort course on AI platforms and ethics that closed applications on August 30, 2025, confirming the programme is a real, standing government training product rather than a one-off announcement (MCIT).
What the protocol does not do is authorize AI to draft, review, or influence an actual ruling or legal opinion. Reporting on the signing found no specified use-cases, budget, or implementation timeline for AI inside case work — the entire visible commitment is training (Middle East Observer). Khattab's own framing of the goal is explicit about this boundary: the programme aims at "developing a deeper understanding of [AI's] capabilities and limitations" so counsellors can "deal responsibly with AI applications," not at putting a model in the drafting chair (ITWeb Africa).
The Fair Case for Keeping AI Out of Courts Entirely
Before endorsing any AI-adjacent training for people who litigate on the state's behalf, it's worth taking seriously the strongest version of the skeptical case. Litigation involves liberty, property, and public money; a hallucinated citation or a misread precedent isn't a minor inconvenience, it's a due-process failure. Large language models are demonstrably unreliable narrators of case law — they fabricate citations with total fluency and no internal signal of doubt. In Mata v. Avianca (2023), a New York federal judge fined two attorneys $5,000 after they filed a ChatGPT-drafted brief citing cases that didn't exist; the judge called parts of it "gibberish" and found the lawyers had acted in "subjective bad faith" once challenged (Wikipedia). That wasn't an isolated incident — legal commentators cataloguing 2025 alone found federal judges sanctioning lawyers $1,500 to $15,000 for AI-hallucinated filings, and even two instances of judges' own draft opinions containing fabricated quotations (Sterne Kessler). A skeptic can reasonably argue that any institutional AI-literacy programme risks normalizing tool use in a domain where the tool's failure mode is confident fabrication, and that training alone doesn't stop a rushed counsellor from trusting an AI-generated summary under deadline pressure.
Why the Order Egypt Chose Is the Right One
That risk is real, but it's an argument for sequencing, not for prohibition — and sequencing is exactly what this protocol gets right. The failure pattern in every sanctioned US case above is the same: a lawyer used a general-purpose chatbot with zero institutional guidance, no verification norm, and no one to ask before filing. Egypt is doing the opposite: building the literacy layer — what these tools can and cannot be trusted to do — as an institutional programme, before any case-facing deployment exists to misuse. Training 600-plus counsellors on AI's failure modes before AI touches a live filing is cheaper, in reputational and legal terms, than discovering those failure modes the way Steven Schwartz and Peter LoDuca did in Mata v. Avianca. A proportionate-regulation instinct says you don't need a comprehensive AI-in-courts statute for a tool nobody has been authorized to use yet; you need exactly what Egypt built — a mandatory competency floor.
The Governance Question Still Deferred
The honest gap is that "no deployment yet" is not a permanent state. The State Lawsuits Authority handles an enormous docket of state contracts and litigation, and the efficiency case for AI-assisted document review or legal research is genuine — which means the day some counsellor uses an AI tool to draft a memo that shapes a real filing is coming, protocol or not. When it does, Egypt will need the second half of what this agreement skipped: disclosure rules when AI assists a filing, a verification requirement before citation, and an audit trail distinguishing AI-assisted work from AI-authored work. None of that needs to exist today. But the Authority and NTI should be building it now, on the same clock as the second Ambassadors cohort, rather than waiting for their own version of a $5,000 lesson.