Executive Summary
This paper addresses the legal and procedural risks arising from the Egyptian Public Prosecution’s announcement of the integration of artificial intelligence (AI) technologies into its work, as part of what it calls “Smart Prosecution.”
The paper does not seek to reject the technology in principle, but rather to deconstruct the official discourse that presents such integration as a neutral form of modernization. It also reveals that it may reshape the procedural power balance in favor of the prosecution, at the expense of the right to defense, the presumption of innocence, and the guarantees of a fair trial.
The paper analyzes five interconnected axes; The first is the Egyptian legislative vacuum, which lacks any legal framework regulating the use of automated tools in investigation and procedural decision-making. The second is the classification of potential AI functions within the Public Prosecution by risk level, ranging from low-risk administrative functions to those that directly affect freedom.
Third, the phenomenon of “disguised automated decision-making,” where algorithmic recommendations become an implicit institutional standard, replacing independent human judgment. Fourth, the risks of structural bias inherent in law enforcement data from which these systems will learn. Fifth and last, deconstructing the five official justifications in the discourse of “smart prosecution.”
The paper concludes that the use of AI in areas affecting personal freedom before a verdict should not be judged solely by speed or accuracy, but by its effect on litigants’ ability to examine, review, and challenge decisions, and on the balance between prosecutorial power and defense rights. The paper also presents a framework of legislative, institutional, procedural, and oversight safeguards that must precede any actual integration.
Introduction
In February 2026, the Egyptian Public Prosecution announced a project to integrate AI-enhanced systems into the justice system, as part of Egypt Vision 2030 and its digital transformation strategy. The official statement presented this path as a transition towards a “smart prosecution” that is capable of analyzing big data, increasing the accuracy and speed of procedures, and enhancing transparency.
This announcement followed the signing of a cooperation protocol between the Ministry of Communications and Information Technology and the Public Prosecution, which includes expanding digital services, integrating databases, and improving decision-making efficiency through shared data repositories.
However, in their current form, such actions do not provide a legal guarantee. They leave fundamental issues undefined, such as the functions these systems will perform within the daily work of the Public Prosecution, the areas where they will intervene within the procedural process, and the limits preventing them from becoming a tool for expanding the power of the prosecution at the expense of the right to defense and personal freedom.
Furthermore, there is a lack of clarity regarding the legal framework governing this use, given the silence of Egyptian legislation on the matter.
This paper does not oppose the technology per se, but it refuses to present it as a neutral modernization. It proceeds on the premise that introducing AI technologies into the work of the Public Prosecution raises procedural questions regarding the right to defense and oversight of discretionary power.
These technologies will be introduced into an institutional structure that has already faced human rights criticism for the broad powers of the prosecution during the investigation and pretrial detention phases, and for the resulting disparity between the prosecution’s tools and the defense’s.
Hence, the use of speed, analysis, correlation, and even prediction tools by the prosecution alone could exacerbate procedural problems unless these tools are accompanied by safeguards that guarantee transparency, accountability, the right to access the reasons for decisions, and the right to appeal and effectively challenge them.
This paper adopts a legal-procedural analysis methodology that combines three approaches: first, a textual reading of the official statement and the cooperation protocol; second, an analysis of the governing Egyptian and international legal frameworks; and third, drawing on comparative experiences with the use of AI in criminal justice systems worldwide.
First: The Legal and Institutional Context
The Egyptian Constitutional and Legislative Framework
The 2014 Egyptian Constitution, as amended, includes several provisions that directly address integrating AI into the Public Prosecution’s work. Article 54 guarantees personal freedom and stipulates that, except in cases of flagrante delicto, no one may be arrested, searched, detained, or have their freedom restricted except by a reasoned judicial order.
It also establishes procedural guarantees related to reporting, contacting a lawyer, being brought before the investigating authority within 24 hours, and the right to appeal, while leaving the regulation of pretrial detention, its duration, and its grounds to the law.
Article 96 stipulates that the accused is innocent until proven guilty in a fair and legal trial that guarantees their right to defend themselves. While Article 98 protects the right to defense and legal representation. Article 99 stipulates that any infringement upon personal liberty or other public rights and freedoms guaranteed by the Constitution and the law is a crime for which no statute of limitations applies.
These constitutional provisions were not designed in a context that anticipated the introduction of automated tools into procedural decision-making. However, the principles they protect, such as the inviolability of personal liberty, the presumption of innocence, the right to an effective defense, and the requirement that decisions affecting liberty be justified by human reason, subject any technological integration to a rigorous constitutional test.
Therefore, the contribution of algorithms to forming suspicion, favoring detention, or constructing the prosecution’s narrative raises a constitutional question regarding the extent to which the decision remains based on independent human judgment, allowing the defense to challenge and appeal it.
At the legislative level, the Egyptian Code of Criminal Procedure does not contain explicit provisions regulating the use of automated tools, algorithms, or artificial intelligence systems in the investigation phases or in the formulation of procedural decisions. The law assigns these procedures to the Public Prosecution, its members, and the competent judicial authorities, in accordance with the conditions and procedures it specifies.
This legislative silence raises a question of legitimacy, particularly if analytical, classificatory, or predictive tools are used in decisions affecting personal freedom, as this use is not based on explicit legislative regulation that defines its scope, safeguards, and mechanisms for review and appeal.
As for the Personal Data Protection Law No. 151 of 2020, it represents an important legislative step; however, it contains significant exceptions. Article 3 excludes from the scope of the law, among other things, personal data related to judicial records, investigations, and lawsuits, as well as data held by national security agencies. Thus, data processing within the Public Prosecution, limited to its connection with investigations and lawsuits, may remain outside the direct protection stipulated by law.
Furthermore, the executive regulations, issued in late 2025 after a long delay, do not, on their own, resolve the lack of detailed regulations regarding criminal data and its analytical use within the justice system. Therefore, it can be said that the general legislative framework for data protection in Egypt still leaves a clear gap regarding the collection, processing, and analysis of criminal data within the justice system.
On the other hand, the Cybercrime Law No. 175 of 2018 allows investigative authorities, under reasoned warrants, to access a wide range of user data and digital communications held by service providers. It also requires these providers to retain records and data for 180 days and provides national security agencies with extensive technical capabilities.
This law has already expanded the legal framework for enforcing laws governing digital data. Therefore, combining it with sophisticated analytics tools capable of linking and classifying data could increase the risks to privacy and freedom of expression.
Second: Statement from the Public Prosecution
The Public Prosecution in Egypt is not solely an accusing body. It rather legally undertakes the functions of investigation, as well as initiating and conducting criminal proceedings. It oversees prisons and other facilities where criminal sentences are carried out, and operates within a hierarchical structure headed by the Attorney General.
This consolidation of responsibilities within a single entity necessitates the introduction of more sophisticated analytical tools, especially compared to systems that institutionally separate investigation and prosecution.
Here, the same institution possesses the authority to assess and evaluate evidence, formulate charges, and make certain decisions restricting freedom during the investigation phase, such as pretrial detention.
Hence, any tool that enhances the efficiency of these processes may, in practice, strengthen the prosecution’s power without necessarily being matched by a similar strengthening of the defense’s power.
The reality of pretrial detention in Egypt gives this discussion a pressing practical dimension. Numerous Egyptian and international human rights reports have documented widespread patterns of the expanded use of pretrial detention, including the continued detention of defendants through the “rotation” of individuals in new cases or on similar charges after their release orders have been issued or after they have reached the legal limits for detention.
In this context, the issue of AI raises additional concerns because any system capable of linking databases, analyzing patterns, and assessing risk could facilitate these practices or lend them an air of technical objectivity, without eliminating the underlying legal and human rights problems.
When the Public Prosecution announced the strengthening of its technological infrastructure with advanced AI systems as part of the modernization of the justice system, its statement used the terminology of technological modernization without specifying the details of its practical application.
The statement used general terms such as big data analysis, improving the accuracy and speed of procedures, enhancing transparency and litigants’ rights, and increasing public trust. However, these terms leave a definitional gap. The statement doesn’t specify what functions these systems will actually perform, where they will intervene in the procedural process, or the limitations imposed on them.
The cooperation protocol with the Ministry of Communications adds another dimension, as it includes database integration, expanded digital services, and the creation of shared data repositories. In practice, this means the project is not limited to improving internal work tools; it constitutes building an infrastructure for collecting, linking, and analyzing data across multiple institutions.
The difference between the two is fundamental. The former may remain within the realm of administrative organization. The latter addresses privacy and expands the state’s ability to build personal and behavioral profiles.
Third: Analyzing Potential Functions and Levels of Risk
The most practical approach to dealing with the “Smart Prosecution” project does not begin with technology as an abstract concept. It is rather the functions these systems might perform within the daily workflow of the prosecution, and the classification of these functions according to their impact on rights.
Not all AI uses are equally risky. Hence, grouping them in a single discourse leads to presenting the project as a simple administrative improvement, all while opening the door to including functions that strike at the very heart of freedom.
Low-Risk Administrative Uses
This category includes digital case and document management systems, internal search engines for retrieving precedents and memoranda, and tools to improve workflow and file distribution. These functions may remain administrative in their impact if they are limited to organizational tasks.
However, the line between them, on one side, and more serious functions on the other side, is not always clear. For example, a file summarization tool appears administrative in its title. Still, in practice, it rearranges the elements of a case and determines what is considered essential and what is relegated to the margins.
Summarizing is selective, and selection carries an implicit judgment on the facts. When the automated summary becomes the lens through which a prosecutor views a case, the initial conviction may shift without anyone realizing it.
The Three Gateways to High Procedural Risks
Gateway One: Case Sorting and Prioritization
When a system is granted the authority to prioritize cases according to pre-existing patterns or indicators derived from accumulated data, it redistributes attention and resources within the prosecution. Determining which report to investigate first, which case to prioritize, and which person to classify as requiring stricter measures—all of this effectively becomes criminal policy.
The difference is that this policy is not passed through legislation. It is not subject to parliamentary oversight either, and it cannot be challenged in court. All of this is because it is embedded in an algorithm whose workings are unknown to anyone outside the system.
This shift from declared criminal policy to embedded algorithms is highlighted by the European Charter of Ethics for the Use of Artificial Intelligence in Justice Systems (CEPEJ). It warns that transferring criminal policy to an internal technological structure, away from public oversight and societal debate, undermines the principles of transparency and accountability.
Gateway Two: Data linking and integration
The cooperation protocol between the Ministry of Communications and the Public Prosecution includes integrating databases and establishing shared repositories. This type of integration, when combined with analytical and predictive tools, expands the state’s ability to build comprehensive profiles of individuals by combining criminal, financial, communication, and geographic records.
At this point, AI transitions from an organizational tool to a linking and monitoring structure. A tool that infringes on privacy and redefines what constitutes “relevant information” within an investigation.
Gateway Three: Assessment and Preference
Any system that produces a degree of risk or probability or a standard recommendation related to a procedural course of action, such as detention or release, expansion or dismissal of the investigation, indictment or amendment,
enters the core of procedural authority.
The central risk is not the explicit delegation of decision-making power to a machine; such delegation is rare. The risk lies in the recommendation becoming a de facto decision within an institution operating under time pressure, productivity assessments, and rigid hierarchies.
The report of the UN Special Rapporteur on the Independence of Judges and Lawyers on AI in justice systems emphasizes that the introduction of AI into justice systems requires rigorous risk assessment and human rights-based management, not solely on efficiency criteria. The report also warns that managing these systems with the logic of technological promises could undermine guarantees of a fair trial and the independence of decision-making.
Fourth: Procedural Risks
The integration of AI does not add a purely technical risk to the structure of justice. It can burden the most sensitive procedural areas, where the prosecution’s narrative is formed before the defense has a chance to confront it, and an initial conviction may be difficult to dismantle later.
Forming and Expanding Suspicion
What is presented under the guise of “big data analysis” and “database linking” may, in practice, become an expanded mechanism for generating suspicion.
Classification and profiling systems do not detect crime impartially. They redirect suspicion according to patterns in the available data. This expansion of suspicion may automatically broaden the circle of those who go through law enforcement procedures, with a direct impact on privacy, freedom of expression, and the right to organize.
Pre-Judgment Decisions
The impact of AI is amplified in the pre-judgment phase. This is because any risk assessment or probability estimation can be directly translated into a decision that jeopardizes the freedom of an individual whose guilt has not been proven.
In the Egyptian context, where human rights reports have documented patterns of expanding pretrial detention, accelerating decision-making through intelligent systems, and presenting it with an “objective” facade, increases the likelihood that the exception will become a more systematic and severe practice.
Formulating the Charge and Constructing the Narrative of Facts
Systems that summarize documents, classify facts, or link data from several sources do not simply present the case file as is. Every summary rearranges the elements of the case. It identifies what appears central and marginalizes others. It may produce a seemingly coherent narrative; however, it omits elements that could be crucial to the defense.
At this stage, the right to a defense is harmed in two ways: First, the accusation is amplified in a way that appears mechanically logical. Second, the center of gravity shifts from the original case file to outputs that the defense lacks the tools to understand or challenge.
Appeals and Review
The appeal and review process becomes complicated when AI contributes to decision-making. Part of the logic of the decision is transferred to a layer obscured by institutional secrecy, commercial design confidentiality, or a technical inability to explain it.
The result is that the appeal remains available in form but not in substance. This is because the defense is faced with a result. And the defense cannot access the premises of such a result.
The Disguised Automated Decision
The most common formula to dispel fears about AI in justice is that “the system supports but does not decide.” Another one is that “the final decision remains human.”
However, this formula ignores how institutions actually operate. In any organization operating under time pressure and within a rigid hierarchy, automated recommendations gradually transform from additional information into a practical standard of correctness, and then into an implicit reference point, the violation of which is punishable.
Numerous factors can combine to exacerbate the risk of disguised automated decisions:
First: Time pressure and file overload.
When an automated summary, classification, or severity rating is presented, these outputs may, in practice, function more as a time-saving measure than as simply additional information. However, this shortcut is not necessarily neutral, as it may narrow the scope of human review, which is supposed to be one of the safeguards protecting the rights of the accused.
Second: Shifting the burden of justification.
What is known as “reliance bias” arises when automated outputs are presented as more objective or accurate than human assessment. In such cases, disregarding the recommendation may become more costly than following it. This happens because rejecting the recommendation requires additional justification, while adhering to it is considered the safer course.
In turn, the burden of justification is effectively reversed. Instead of the organization proving the recommendation’s validity, the individual must justify why they disregarded it.
Fifth: Unequal Impact on Specific Groups
The risks of AI are not distributed equally among all groups in the justice system.
Some groups bear the heaviest burden because they are already overrepresented in law enforcement data. It may be as well because the nature of their activities makes them more susceptible to algorithmic classifications that broaden suspicion against them.
- Political opponents and activists:In environments where defendants are repeatedly “rotated” through similar cases with recurring charges, systems that link cases and analyze patterns can automatically generate new suspicions based on past records and social networks. Decisions that once required intentional human judgment may instead emerge as seemingly neutral, automated outcomes.
- Journalists: Under broad definitions of counterterrorism and cybercrime laws, data analytics can expand targeting. It could include anyone who has contacted a designated source or published content that intersects with “suspicious patterns” extracted by the algorithm.
- The poor and economically marginalized: This group is overrepresented in law enforcement data everywhere. Consequently, any system that learns from this data will consider them “higher risk” because they are recorded more frequently. This is not necessarily because they are more likely to commit crimes. It is a structural discrimination that perpetuates itself.
- Women in “Public Morality” Cases: These are cases of “violating Egyptian family values” and public morality, where social media posts are often used as evidence. It is expected that automated analysis systems will enable more expansive tracking and categorization. This, in turn, heightens the risk of prosecution in matters that fundamentally concern freedom of expression and the right to privacy.
Sixth: Biased Data
The bias cycle is a closed loop. When an algorithm is fed historical data reflecting selective targeting of specific groups or regions, it, in turn, produces estimates that treat these groups or regions as higher risk. This leads to directing more resources and attention to them. Eventually, this increases their presence in the records.
The system then learns from this artificial density that it is further evidence of risk. Thus, the algorithm does not reveal objective reality so much as it reuses and recycles previous targeting patterns, giving them a controlled, technical appearance.
United Nations reports on racism and artificial intelligence highlight that analytical models can replicate entrenched forms of discrimination, since historical biases are carried forward through the data used to train these systems.
Comparative literature on risk assessment tools also indicates that these tools often rely on biased data. At the same time, institutions tend to treat their outputs as more accurate and objective than they actually are.
One of the most dangerous consequences of this cycle is that it alters the very function of the criminal justice system. Instead of holding someone accountable for a specific act based on evidence related to that act, the focus shifts to their position within a statistical pattern or probabilistic classification.
At this point, the presumption of innocence is effectively eroded in the pre-trial stage, as suspicion arises from data processing rather than from specific individual facts.
These phenomena are not theoretical, and the following comparative experiments clearly demonstrate them:
The American Experience: COMPAS and the Loomis Case:
In the United States, software is sometimes used to estimate a defendant’s future re-offending likelihood, also known as a “recidivism risk assessment.” The most well-known of these programs is the COMPAS system, a privately owned system that processes a defendant’s data to produce a risk score or classification.
In the State vs. Loomis case, heard by the Wisconsin Supreme Court in 2016,
this system was used among the materials reviewed by the court when sentencing the defendant. The defendant objected to this because the system was not transparent. The defendant could not know how the program arrived at its assessment, examine its accuracy, or effectively challenge its reasoning. All this is due to its confidential commercial nature.
However, the court accepted its use in principle, but stated that it must remain subject to clear warnings and limitations. The court also emphasized that its outputs should not be treated as the sole determining factor in deciding the sentence or length of imprisonment.
In the same context, a ProPublica investigation concluded that the system makes racially discriminatory errors, being more likely to classify non-recidivist Black defendants into higher risk categories compared to white defendants.
This experience reveals the risks of structural bias, the difficulty of effective appeals when the logic of the tool is obscured, and the potential for “support” to become a profoundly influential factor in judicial decisions.
The Dutch Experience: SyRI
In February 2020, the International Court of Justice (ICJ) ruled that the legal framework of the SyRI system was incompatible with Article 8 of the European Convention on Human Rights.
The system was used to detect patterns of fraud, including fraud in social welfare and benefits, by linking and analyzing multiple government databases. The court ruled that the legislation governing the system, in its existing form, did not strike a fair balance between the public interest and the right to privacy, and that the transparency and safeguards requirements were insufficient given the extent of the intrusion into privacy.
If this standard has been applied in the context of social welfare, its relevance becomes even more pressing in law enforcement, where personal liberty may be directly affected.
Seventh: Deconstructing Official Justifications
The official discourse of “smart prosecution” is structured around 5 arguments. Each one contains a semblance of plausibility, but at the same time, it is tangled and obstructed under the scrutiny of rights principles.
The Complexity of Crimes Necessitates Smart Tools
This argument seems logical if read as a purely technical observation. However, it becomes problematic when used as a license to expand power without restraint.
What complexity requires is raising the quality of human investigation, strengthening expertise, and protecting the independence of decision-making. It does not require introducing an algorithmic weighting layer that produces a quicker suspicion and widens the gap between the prosecution and the defense.
The Argument for Expedited Justice
No one disputes the need to minimize undue delays. However, expedited justice is not about reducing time at any cost. Speed within a system that is unbalanced between prosecution and defense does not produce faster justice. It rather presents a greater capacity to restrict freedom before reaching a verdict.
General Comment No. 32 of the Human Rights Committee affirms that a fair trial
requires sufficient time and appropriate facilities for the preparation of the defense. It also states that equal opportunity requires practical equality between the parties.
Introducing an expedited system that does not provide the defense with equivalent knowledge, nor does it enable challenging its effect, produces more efficient injustice.
The Argument for Accuracy and Reducing Human Error
This argument is based on a misleading comparison between “human errors and a more accurate machine.” Legal accuracy is not merely statistical accuracy.
It is accuracy that can be reasoned, understood, and challenged.
Human error can be addressed because its logic is understandable and its path is traceable. However, automated errors become more dangerous when they are presented as “objective” facts and the logic behind their creation is concealed.
The Argument for Transparency and Building Trust
This argument is the most contradictory to reality. The very statement that promises transparency does not disclose the implementing companies, the system specifications, or even the protection standards.
True transparency entails enabling the litigant to know that an algorithm contributed to a decision concerning them, and to challenge its impact.
Without this, transparency becomes internal transparency between government entities, with no regard to transparency towards the rights holder.
The Argument for Leadership and Digital Transformation
Linking the project to Egypt Vision 2030, prioritizes justice as a competitive arena.
Technological leadership does not equate to legal leadership. Introducing data analysis and integration systems within an institution possessing coercive powers, without disclosing rules of use, data limits, and oversight mechanisms, only expands the state’s executive capacity under the guise of modernization.
It is worth noting here that the European Artificial Intelligence Act (AI Act) of 2024 classifies the use of AI in law enforcement and criminal justice as “high risk”. It requires mandatory impact assessments, transparency, interpretability, and rigorous human oversight.
This classification represents an international institutional recognition that the field of criminal justice requires exceptional safeguards, not just general promises.
Eighth: Legal Responsibility and Social Gaps
One of the central problems is the lack of a clear definition of responsibility when an algorithmic recommendation leads to a violation of a person’s rights. Despite the simplicity of this issue in principle, the current Egyptian legal framework does not provide a specific answer.
If a prosecutor relies on biased data in deciding on pretrial detention, and it later becomes clear that the decision was unjustified, the issue of legal responsibility remains ambiguously distributed among several parties.
The decision may be attributed to the prosecutor, even though they were unaware of the logic that produced the classification. The state, as the operator of the system, may also be held responsible, as may the developing company, given the possibility of design flaws or data bias.
This ambiguity may increase in the absence of an impact record documenting the algorithm’s role and its actual influence on the decision. The absence of this definition leads to a serious accountability gap. The victim often lacks the means to prove that the algorithm contributed to the decision issued against them, because they may not even be aware of its use.
Conversely, the prosecutor may rely on the automated recommendation to justify their decision, while the developing company hides behind trade secrets. Meanwhile, the state maintains that the final decision remained a human one.
Thus, responsibility is distributed in a way that practically leads to its disappearance, while the victim alone bears the consequences. An analysis of accountability is incomplete without addressing the identity of the implementing entities and their role within this system.
The issue extends to the companies developing these systems for the Egyptian Public Prosecution. Questions regarding such companies include whether they are local or international, their human rights record, and the extent to which their contracts include clear commitments to transparency and independent auditing. It also extends to the impact of intellectual property rights in shielding algorithms from scrutiny. This shielding may prevent algorithms from being subject to effective review in contexts that touch on freedom and fundamental rights.
The digital transformation of justice also creates a new divide, between those who have the tools to confront the technology and those who do not. If part of the decisions of the prosecution is based on algorithmic outputs, effective appeals
will require specialized technical expertise, such as understanding how algorithms work, detecting bias in data, and using independent experts to evaluate outputs.
While large law firms may be able to utilize this expertise, individual lawyers
in remote areas will not. This means that the digital transformation of the prosecution service may deepen inequality in access to effective justice. All in turn means that wealthier litigants will have better tools to fight, while the poor remain facing a more complex system with weaker tools.
This gap violates the principle of equality before the law, guaranteed by both the Egyptian Constitution and the International Covenant on Civil and Political Rights.
Modernization Begins with Rights
What this paper reveals is not about a technological project whose standards of efficiency and modernization can be evaluated. It is rather a transformation in the structure of procedural power being implemented under the guise of neutral language.
There is one question the paper could not answer, because it transcends its methodological scope, but it remains present, in the background of the entire analysis: Can a framework of safeguards, however theoretically coherent, actually function in a way that does not violate rights and uphold justice?
Perhaps the most important contribution this discussion makes is a re-examination of an assumption recurring in the discourse of modernization, namely, that the challenges to justice can be addressed primarily at the level of tools. However, the analysis presented in the paper indicates that the impact of technology is not determined solely by its characteristics, but is also shaped by the institution that uses it and the legal and procedural framework within which it operates.
Therefore, introducing more efficient tools into a system already suffering from problems may exacerbate some existing imbalances or make them less apparent. For this reason, it is difficult to separate the discussion of AI in the context of justice from the broader discussion of criminal justice reform.
Furthermore, isolating the technical issue from its institutional and legal context
may lead to an incomplete understanding and to recommendations of limited effectiveness.
The paper also raises questions that require separate attention, including: How do Egyptian lawyers actually interact with the digital transformation in justice facilities, and what impact does this have on their ability to defend?
What are the technical details of the systems being developed or acquired, and can they be subjected to independent auditing?
How is criminal policy actually shaped when it moves from declared human decisions to embedded algorithms, and what oversight mechanisms are possible for this transition?
What is the long-term impact on the structure of the legal profession and the independence of prosecutors themselves when part of their professional judgment becomes subject to invisible technological calibration?
Finally, this paper is not addressed solely to researchers, but to all those who have a say or influence in this matter, such as legislators, members of parliament, judges, lawyers, and civil society organizations. All parties should treat technology as not an inevitable fate, and legislative silence does not imply consent. Modernization that does not begin with rights ends up modernizing tools of control, not tools of justice.