One Chance to Challenge a Red Notice: Why You Should Not Rely on AI When Applying to the CCF

Challenging a Red Notice is a complex process that involves dealing with different legal systems and building a strategy within the Commission’s limited mandate. What makes it even harder is that an applicant gets only one attempt before the Commission for the Control of INTERPOL’s Files (CCF). If a case is lost because of weak arguments, fabricated references to non-existent case law or legal provisions, or factual errors, it is almost impossible to put this right in a subsequent application.

More and more often, a person subject to a Red Notice or a diffusion decides to save on legal fees and asks artificial intelligence to draft the complaint to the Commission. In most cases the text may look convincing, but the client cannot assess it against the Commission’s practice, verify whether the references to legal provisions are accurate, or judge whether a particular document should be submitted at all.

In this article I explain why the cost of saving on a lawyer is particularly high in CCF proceedings, and why you should not rely on artificial intelligence without professional review.

Why there is no room for error at the CCF

There is no appeal at the CCF in the classical sense. The Commission’s decisions are final, and the only mechanism for reviewing them is a request for revision under Article 42 of the CCF Statute.

The conditions for revision are quite narrow: it requires new facts that were unknown to the applicant when the original request was examined and that, had they been known, would likely have led to a different conclusion by the Commission. Such a request must be filed within six months of the discovery of those facts.

The Commission states expressly that the purpose of a revision request is not to re-assess arguments, but to assess new circumstances and documents in terms of their novelty and their significance for the conclusions the Commission reached in its decision. A repeat complaint that essentially duplicates the previous one and contains no new facts is therefore declared inadmissible.

In practice, revision requests are extremely difficult to work with. On the one hand, the criteria are clear; on the other, the scope for assessing new facts is limited, which means the Commission sets a high bar for the quality of those facts. The task is further complicated by the fact that if the original submission lacked a strong legal position, that position cannot be added later. In other words, new facts and evidence must tie in with arguments that have already been submitted, rather than create new ones.

A document that was in your possession but was not attached to the complaint is not a new fact either. Revision is possible only in respect of circumstances that were genuinely unavailable, not ones that were simply not submitted. That is why analysing the available evidence at the stage of the initial submission is crucial. AI typically suggests submitting every document, yet some of them are not always relevant to the Commission or may even harm the applicant’s position.

A lost complaint can mean years of living with travel restrictions, since the time it currently takes to process a deletion request often significantly exceeds the time limits set in the CCF Statute.

AI exaggerates the circumstances and invents violations

Recently a client came to me after the CCF had refused to delete a diffusion against him. It was an ordinary economic case, with court decisions in the client’s favour and a refusal of extradition already on file. The client openly admitted that he had written the CCF complaint himself using AI.

An analysis of the case file and the CCF’s decision showed that some of the arguments had been distorted and others were entirely inappropriate. AI exaggerates the circumstances in an attempt to prove violations that, in substance, do not exist. Very often this concerns arguments about violations of Articles 2 and 3 of INTERPOL’s Constitution. AI supports claims of a predominantly political character or of serious human rights violations with general country reports or low-quality facts, without a systematic analysis under the predominance test. Sometimes such arguments become the centrepiece of the complaint, which makes it a losing one from the outset. At times AI argues the applicant’s innocence or the inadmissibility of the prosecution’s evidence as a breach of INTERPOL’s rules, which is entirely wrong.

In this case we were unable to help the client, as he had no new facts that would allow a request for revision to be filed. It is a striking illustration of the consequences of relying on AI and saving on a lawyer.

Trust AI, but verify

According to the database compiled by lawyer and data researcher Damien Charlotin, there are already more than 2,200 cases worldwide in which a court or tribunal has identified AI “hallucinations” in procedural documents. As of 26 October 2026, 1,514 of them came from the United States, 234 from Canada, 112 from Australia and 76 from the United Kingdom. The pace tells its own story: when LexisNexis and Fortune reported on this database in May 2026, it held just over 1,300 cases. In most of them, ordinary publicly available chatbots were involved.

There are no publicly known cases at the CCF so far, and that is no coincidence. The Commission’s decisions are published only as anonymised excerpts, and an unsuccessful complaint does not make the news. We simply do not know how many applicants have already been refused because of an AI-generated text. But the logic of all the cases above is the same, and there is every reason to expect it to work the same way at the CCF.

At INTERPOL’s 93rd General Assembly in 2025, amendments to the CCF Statute were adopted, including on abuse of the CCF procedure. The new rules empower the Commission to take measures in response to improper conduct and abuse of its procedure. The Commission may now formally reject a request if it considers that its procedure has been abused. A fabricated CCF decision or AI-generated evidence in a complaint may therefore have real legal consequences for the applicant.

Why the risk of using AI is even higher in INTERPOL cases than in court

In ordinary court proceedings, a fabricated precedent or flawed reasoning may be spotted by the opposing party or the court, and you will have an opportunity to correct yourself or amend your statement of claim. In cases before the Commission for the Control of INTERPOL’s Files the stakes are higher, for several reasons.

First, you will not know what problems your complaint has until you receive the Commission’s final decision. The Commission may, of course, hold additional consultations with the applicant on the subject of the complaint, but this happens extremely rarely and usually concerns technical matters — for example, a document that was sent is damaged or of poor quality, or the Commission asks for permission to disclose part of the arguments to the country that requested the Red Notice. Moreover, you will not know the content of the data source country’s objections until you receive the final decision.

The second significant reason is that AI knows almost nothing about CCF practice because of its closed nature. In other words, unless you have dozens of completed CCF cases behind you, you cannot train your AI agent to build a complaint on the Commission’s practice, and this will inevitably result in a low-quality submission. Some of the Commission’s decisions are published in anonymised form on INTERPOL’s website, but their number is quite limited. For example, only 4 decisions were published for 2025 and 5 for 2024. By comparison, according to the Commission’s official report, 539 decisions on the merits of deletion requests were adopted in 2024. That is why publicly available practice is not representative. In this situation, when AI has no data to compare against, it may invent data, including citing Commission practice from decisions that do not exist.

The two problems above lead to the choice of a wrong strategy, which can no longer be changed afterwards. Success at the CCF depends on the choice of strategy and on the evidence that proves the violations you allege. Even more important is the order in which facts and evidence are presented, and what is deliberately included and what is not. AI does not know the position of a particular NCB, the political context or the history of similar cases, and it bears no responsibility for the outcome. That is precisely why lawyers who understand CCF practice are able to build the right strategy.

Conclusion

Challenging a Red Notice or a diffusion before the CCF is a procedure with one real chance. Revision under Article 42 is possible only on the basis of new facts, which most people simply do not have. AI can produce text that looks like a legal document. But it will not check whether the decisions it cites actually exist, it does not know the Commission’s practice, it does not see the risks and it is not accountable for the consequences. If an INTERPOL notice has been published against you, it is better to turn to a lawyer who specialises specifically in CCF procedures, because saving money on the first complaint may lead to irreparable consequences.