STF Before the Mirror: Moraes's Defense Turns the Master Case Into an Accusation Against the Investigator

In a 44-page submission filed on the eve of an extraordinary Supreme Court session, Alexandre de Moraes argues that the investigation ordered by André Mendonça is null, questions the integrity of the material produced and accuses his colleague of politically directing the inquiry. The document contains relevant legal arguments, but also advances conclusions supported by very different levels of evidence.

Fernando Fagundes Ferreira · 2026-09-15

This article was prepared during the extraordinary session of Brazil's Supreme Federal Court (STF) held on 15 September 2026 and will be updated once the trial is concluded or suspended.

The crisis opened at the Supreme Federal Court by the Banco Master case can no longer be understood merely as a controversy over the possible conduct of Alexandre de Moraes. The submission the justice sent to the Court's president, Edson Fachin, on 14 September shifts the centre of the problem by attacking the very legitimacy of the procedure that produced the suspicions against him.

Across 121 numbered points spread over 44 pages, Moraes argues that the investigation was conducted without an adequate procedural basis, without a request from the Office of the Prosecutor General, without authorisation from the full bench and with undue interference by the case's rapporteur, André Mendonça, in the investigative work of the Federal Police. The defence, therefore, does not confine itself to denying facts. It seeks to dismantle the very path through which those facts were produced, organised and presented to the Court.

That shift is decisive because it creates two parallel questions. The first remains the original one: is there concrete evidence indicating irregular action by Alexandre de Moraes to the benefit of Banco Master or Daniel Vorcaro? The second, however, has become institutionally as serious as the first: was the procedure that raised those suspicions conducted by a competent authority, within the rules of the accusatorial system and with minimum guarantees of evidentiary integrity?

It is on this terrain that the Court now judges not only an investigation, but also its own mechanisms for producing evidence, exercising control and assigning responsibility.

The legal core of the defence

The most consistent part of the submission lies in the discussion of jurisdiction and investigative initiative. Moraes states that André Mendonça ordered investigative measures against a Supreme Court justice on his own motion, without a request from the Prosecutor General's Office, without any initiative by the Federal Police and without prior authorisation from the full bench. To sustain this argument, he invokes the constitutional accusatorial system, the Court's original jurisdiction to prosecute its own justices and the prohibition on judicial initiative during the investigative phase.

The argument gains weight because the Prosecutor General's own submission, reproduced in the document, treats the measure as doubly void: on one hand, because the judge would have exceeded the limits of his role in the pre-trial phase; on the other, because any deepening of an investigation against a justice of the Court should have been submitted to the president of the STF and then to the full bench.

This is probably the most sensitive point of the entire controversy, precisely because it does not depend on demonstrating political intent, personal persecution or conspiracy. The question comes earlier: was André Mendonça legally able to do what he did, in the way he did it?

If the answer is no, the problem exists regardless of any assessment of the substance of the suspicions against Moraes.

The submission also recounts a procedurally unusual episode. According to a Federal Police report cited by Moraes, investigators were summoned to Mendonça's chambers and personally handed a physical copy of a still-unsigned decision, while the Prosecutor General's Office had not been formally notified. The decision was later compared with the version made available in the electronic case file.

That episode, on its own, does not prove fraud, persecution or political purpose. But it reinforces the need to examine the regularity of the procedures adopted, above all because the rapporteur's own conduct has become a central part of the dispute.

The difference between a note, a message and evidence

The second important axis of the defence concerns the analysis of the material extracted from Daniel Vorcaro's mobile phone.

Moraes argues that a significant part of the conclusions against him was built from texts found in the device's notes application. The central criticism is methodological: a note written by Vorcaro is not automatically equivalent to a message sent to Alexandre de Moraes, much less to a message produced by the justice himself.

According to the submission, of the 52 notes considered in the report, 47 were not found in any WhatsApp conversation. In the remaining cases, the link was built from the temporal proximity between the creation of the text and certain activity records in the application.

The distinction is fundamental because it involves different evidentiary stages: authorship of the note, actual sending, identity of the recipient, receipt, reading and any reply. Merging these stages into a single inference can create the appearance of communication where only a temporal correlation exists.

The submission also states that there is no message authored by Moraes indicating prior knowledge of a police operation, favourable treatment of Banco Master or legal advice provided to Vorcaro.

This point is relevant, but it does not settle the matter. The document presented by Moraes is a defence pleading and, for that reason, a definitive conclusion would require comparison with the original forensic report, the raw data extracted from the device, the forensic logs and the complete documentation used by the Federal Police.

The strength of the defence lies in showing that certain associations may have been presented as more solid than they really were. Its limit lies in turning that methodological criticism, by itself, into complete proof that no relevant fact exists.

The chain of custody as an institutional problem

Another important block of the submission deals with the chain of custody and with the way the digital material was accessed, analysed and forwarded.

Moraes reproduces internal Federal Police reports pointing to problems related to early access to raw data collections, the concentration of sensitive information and the possibility that analyses might reach the case file without adequate technical review by the responsible chief investigator. One of the passages cited argues that removing that review would allow the analyst's inferences to be forwarded without the methodological control normally exercised by the police authority.

Another passage warns of the risks arising from sending copies of sensitive material outside the police environment, which would increase points of exposure and hinder the traceability required by the chain of custody.

This is perhaps one of the most important aspects of the case because it transcends the personal dispute between the justices. If there were failures in the preservation, analysis or circulation of the material, the problem affects the reliability of the evidence regardless of who is being investigated.

At the same time, the distinction between risk and proof must be maintained. The existence of weaknesses in the chain of custody may justify an audit, a methodological review and challenges to the validity of certain conclusions. It does not automatically authorise the claim that the material was tampered with or fraudulently produced.

Where the defence begins to go beyond what it demonstrates

It is precisely at this point that the submission becomes most vulnerable.

Moraes argues that certain metadata would indicate that the report was opened and finalised on the same day on Federal Police computers. From that, he suggests that the document was not produced entirely within the institution.

The question about the document's origin and flow is legitimate. The problem arises when that doubt is converted into a hypothesis of participation by external agents, possibly foreign, and then associated with an attempt to interfere in Brazil's 2026 elections.

The 44 pages do not demonstrate who those agents would be, which foreign institution would be involved, how the material would have been produced outside the Federal Police or by what mechanism this would have turned into an electoral interference operation.

Metadata can raise suspicions about the origin, flow or editing of a document. It does not automatically identify external authorship, nor does it demonstrate an international operation.

That difference is central because, in some passages, the document mixes verifiable facts, institutional interpretations and broader political conclusions. The risk is that the strength of the former is weakened by the excess of the latter.

The accusations against André Mendonça

The submission also gathers serious accusations about André Mendonça's conduct.

Moraes states that the then rapporteur showed a specific interest in opening an investigation against him and requested, more than once, that members of the Federal Police present some initiative to that effect. This allegation is attributed to an intelligence report reproduced in the document, according to which there would have been differential treatment of authorities in similar situations.

The pleading also mentions interference in target selection, discussion of precautionary measures and participation in negotiations related to plea bargain agreements, including ones involving Alexandre de Moraes and the president of the Senate, Davi Alcolumbre.

These allegations deserve scrutiny because, if confirmed, they may indicate an improper convergence between the functions of judge, investigator and prosecutor.

But the difference between describing conduct and proving purpose must be preserved. Showing that a given meeting took place or that a given question was asked is not the same as proving an intention to pursue a political target. And showing a possible irregularity does not automatically prove an electoral strategy.

The submission itself seeks to build that bridge through chronology.

Chronology as an indication, not as complete proof

Moraes argues that Mendonça already knew, months earlier, of references to his name in documents related to the case, but did not refer the matter to the full bench at the time. According to the defence, the investigation only advanced on the eve of the 7 September demonstrations, which would reveal a political-electoral purpose.

The timeline may be relevant. It may indicate selectivity, political opportunity or a change of behaviour that deserves explanation. But chronology alone does not demonstrate intent.

To turn temporal coincidence into a conclusion about electoral purpose, additional elements would have to be identified, such as communications, coordination, planning or actions deliberately linked to political actors.

It is precisely this kind of distinction that must be preserved in analyses of lawfare. Procedures can be politically instrumentalised, but that instrumentalisation must be demonstrated, not merely inferred from the proximity between judicial acts and political events.

The success of the operation and the leak thesis

Moraes also uses the outcome of the operation itself as a defensive argument.

Daniel Vorcaro was arrested at the airport, carrying a genuine passport and his own mobile phone. The operation therefore achieved its central objective. The submission uses that fact to question the hypothesis that a leak had occurred in order to favour the person under investigation.

The reasoning is intuitively strong because a leak intended to frustrate the arrest seems hardly compatible with a suspect captured in the ordinary way, with no signs of flight and carrying the main device of interest to the investigation.

Even so, the success of the operation does not eliminate every possibility of improper transmission of information. A leak may have been partial, poorly understood or incapable of changing the final outcome.

For that reason, the success of the arrest weakens a particular narrative about the leak, but does not in itself prove that no leak occurred.

The 7,742 documents and the problem of scale

Another point explored by the defence is the volume of material analysed.

Moraes states that 7,742 documents spread across dozens of proceedings reveal no contact of his with the authorities responsible for the operation, no prior knowledge of the arrest and no action taken to benefit Vorcaro.

The figure is substantial, but it should not work as a self-sufficient argument. More important than the total number of documents is whether that set corresponds to the relevant universe and whether the facts being sought would necessarily leave traces in that material.

The submission, however, adds a more objective claim: Moraes states that he has never ruled on any proceeding involving Banco Master or Daniel Vorcaro and argues that no official act of his own is directly related to those parties.

That statement has the advantage of being verifiable independently of either side's narrative.

The central paradox of the document

Perhaps the most interesting aspect of the submission lies in Moraes's criticism of the method used against him.

The defence argues that the report started from a previously formulated hypothesis and selected fragments capable of confirming that hypothesis. According to Moraes, the material was organised so as to turn correlations and assumptions into an incriminating narrative.

The problem is that, in some passages, the response itself runs a similar risk. Concrete procedural facts come to be incorporated into a broader narrative of political persecution, electoral interference, foreign participation, revenge and continuity of the 2023 attempt at institutional rupture.

This does not make the investigation and the defence equivalent. But it reveals something institutionally deeper: each side has come to question not only the other's conclusions, but the very method used to produce them.

That is where the crisis ceases to be a dispute about isolated facts and becomes a dispute about legitimacy.

Institutional autophagy

The Court no longer has only to decide whether Alexandre de Moraes should be investigated. It must also assess whether the procedure used to investigate him respected institutional competences, the limits of judicial action and the minimum guarantees of evidentiary integrity.

At the same time, it will have to do so in an environment in which members of the Court itself appear as the person investigated, the investigator, the critic of the investigator, the authority responsible for controlling the process and, finally, the judges of the conflict.

Institutional autophagy thus acquires a deeper dimension. The problem is not only a justice judging a justice, but an institution obliged to show that it can produce a decision perceived as legitimate when its own members are already embedded in the dispute, directly or indirectly.

This is also why the case matters for the study of lawfare. The instrumentalisation of law does not necessarily begin with a conviction. It can emerge in the choice of target, in the way evidence is produced, in the control of the investigation, in the selection of the material disclosed, in the public construction of the narrative and in the definition of who is legitimate to investigate whom.

In that context, the central question ceases to be simply which justice is right and becomes whether the same rules still apply when those who must enforce them are also parties to the conflict.

The STF before the mirror

Alexandre de Moraes's submission brings together consistent legal arguments, accusations that need to be examined and conclusions that go beyond what the document itself sufficiently demonstrates. Treating everything as equally proven would be a mistake, just as dismissing it entirely as mere defensive rhetoric would be.

If André Mendonça exceeded his competences and assumed investigative functions that were not his to assume, that is serious regardless of whether there is evidence against Moraes. If there is legitimate evidence against Moraes, the irregularity of a given procedure should not automatically turn verifiable facts into non-existent ones.

The most delicate institutional point arises when the validity of the rules comes to depend on who occupies each position in the conflict. When that happens, the Court no longer faces only a concrete case and begins to face a question about itself: its capacity to preserve criteria of competence, impartiality and control that are recognised as valid for everyone.

That is the test this Tuesday's session places before the Court. It is not only a matter of finding out which version will prevail, but of knowing whether there is still a rule capable of surviving the dispute between those who should guarantee it.

Because, at this stage of the crisis, no one needs to win for everyone to lose.

Update in progress: this article was prepared during the extraordinary STF session on PET 16.662. The text will be updated after the full bench's decision, any request for further review, a suspension of the trial or any new element capable of altering the analysis presented here.