The debate around the proposed CSA Regulation is often dominated by what people fear it might do. We wish to leave less space to myths, more to facts.
Beyond the myths
#1
“The CSA Regulation will allow mass surveillance and read everyone’s messages. It will allow for your chats to be controlled.”
The facts: The proposed legislation only foresees detection of child sexual abuse material and grooming patterns, through a structured process.
Before any detection can take place, online service providers would need to complete a risk assessment, implement preventive measures and, if a significant risk remains, obtain a time-limited detection order from an independent judicial or administrative authority.
A computer processes data according to rules and algorithms. It does not “read”, interpret, understand, or store messages. It looks for patterns based on what it is trained on. Technologies used to detect child sexual abuse material and grooming pattern are built and trained exclusively to understand and recognise that type of information. A As a metal detector can detect only metal and cannot identify or flag anything else that may be buried underground, child sexual abuse material detection technologies can only recognise child sexual abuse videos or images, and are not able to understand or read any other type of content.
#2
“The Regulation could be used to spy on journalists, activists or political opponents.”
The facts: the proposed Regulation includes multiple safeguards to prevent misuse.
Detection can only be authorised in specific high-risk services, for a limited period, using technologies approved by the independent EU Centre and only after authorisation by a judicial or independent administrative authority.
The Regulation is about protecting children from sexual abuse, not creating new surveillance powers. Repurposing these technologies for indiscriminate monitoring of all communications would be technically impossible, unrealistic and unlawful under the proposed CSA Regulation.
#3
“The Regulation will break end-to-end encryption.”
The facts: the proposal does not require providers to weaken or remove encryption. It does not mandate any specific technology for detecting child sexual abuse. Instead, it sets strict legal requirements that any detection technology must meet, including that it is as accurate as possible, has the least intrusive impact on users’ rights, and respects privacy before it can ever be authorised for use.
This approach allows the law to remain effective as technology evolves. Rather than prescribing a single technical solution, it defines the safeguards that any solution must satisfy. Several privacy-preserving technologies already exist and are used today for purposes such as detecting malware or warning users about harmful content without breaking encryption.
Research shows that around two-thirds of children who receive sexually explicit material online receive it through private messaging, and offenders frequently move conversations to encrypted services to evade detection. Children deserve the same level of protection regardless of the technology used to communicate.
#4
“Detection technologies produce too many false positives.”
The facts: detection technologies have been used for more than a decade and are continuously improved. PhotoDNA, for example, is being used by over 150 organisations globally including service providers (Microsoft), NGOs (e.g. NCMEC) and law enforcement in the EU (e.g. Europol).
Known child sexual abuse material is identified using digital fingerprints (“hashes”), while newer material and grooming are detected using highly trained machine-learning models. Any flagged content is then reviewed by trained specialists before any report is made to law enforcement. To avoid false positives, specific threshold and accuracy requirements will be set by the EU Centre to ensure that high standards are met.
Many technologies, such as speed cameras, have false positives. Nevertheless, societies have opted to deploy them as the objective of reducing road accidents is important enough to accept a low number of false positives where there are no criminal consequences for individuals wrongly flagged. No technology is perfect, but the systems used are designed to achieve extremely high accuracy while proportionate to respond to the scale of the harm.
#5
“The existing Interim Regulation is enough.”
The facts: since 2021, the Interim Regulation has allowed messaging and webmail providers to voluntarily detect, report and remove child sexual abuse material on their platforms. As negotiations on a permanent EU framework are still ongoing, the Regulation was renewed in 2026. However, it provides only a temporary legal basis for some providers to carry out voluntary detection.
It was never intended to be a permanent solution, and it does not create a comprehensive framework for prevention, accountability, oversight or cooperation across the European Union.
The proposed CSA Regulation aims to establish long-term, harmonised rules while maintaining strong safeguards for fundamental rights.
#6
“The Digital Services Act already protects children online.”
The facts: the Digital Services Act is an important piece of legislation, but it has a different purpose.
It improves the safety and accountability of online platforms, but it does not create a dedicated framework for detecting, reporting and removing child sexual abuse material or tackling online grooming. It also does not establish the institutional framework needed to coordinate the fight against online child sexual abuse across the EU.
The proposed CSA Regulation addresses these specific gaps. They would serve different but complementary purposes. The DSA provides broad horizontal rules for online safety, while the CSA Regulation would provide the specialised legal tools needed to combat one of the most serious forms of online crime.
#7
“Detection should only happen when there is already suspicion.”
The facts: in most cases, there is no prior suspicion because abuse is hidden.
Children often do not disclose nor recognise abuse, grooming is a highly manipulative process for victims to recognise, offenders may have no previous criminal record, and child sexual abuse material can circulate for years before anyone becomes aware of it.
The proposed Regulation therefore allows targeted detection only where there is evidence that a service presents a significant risk, and only after independent authorisation and strict safeguards. Without detection, many children would never be identified as victims or safeguarded. This reflects the reality of how online child sexual abuse is committed and detected.