The latest and nearly final version of the proposed UN Cybercrime Convention—dated May 23, 2024 but released June 14—leaves security researchers’ and investigative journalists’ rights perilously unprotected.
The world benefits from people who help us understand how technology works and how it can go wrong. Security researchers, whether independently or within academia or the private sector, perform this important role of safeguarding information technology systems. Relying on the freedom to analyze, test, and discuss IT systems, researchers identify vulnerabilities that can cause major harms if left unchecked. Similarly, investigative journalists and whistleblowers play a crucial role in uncovering and reporting on matters of significant public interest including corruption, misconduct, and systemic vulnerabilities, often at great personal risk.
For decades, EFF has fought for security researchers and journalists, provided legal advice to help them navigate murky criminal laws, and advocated for their right to conduct security research without fear of legal repercussions. We’ve helped researchers when they’ve faced threats for performing or publishing their research, including identifying and disclosing critical vulnerabilities in systems. We’ve seen how vague and overbroad laws on unauthorized access have chilled good-faith security research, threatening those who are trying to keep us safe or report on public interest topics.
Now, just as some governments have individually finally recognized the importance of protecting security researchers’ work, many of the UN convention’s criminalization provisions threaten to spread antiquated and ambiguous language around the world with no meaningful protections for researchers or journalists. If these and other issues are not addressed, the convention poses a global threat to cybersecurity and press freedom, and UN Member States must reject it.
This post will focus on one critical aspect of coders’ rights under the newest released text: the provisions that jeopardize the work of security researchers and investigative journalists. In subsequent posts, Wwe will delve into other aspects of the convention in later posts.
How the Convention Fails to Protect Security Research and Reporting on Public Interest Matters
What Provisions Are We Discussing?
Articles 7 to 11 of the Criminalization Chapter—covering illegal access, illegal interception, interference with electronic data, interference with ICT systems, and misuse of devices—are core cybercrimes of which security researchers often have been accused of such offenses as a result of their work. (In previous drafts of the convention, these were articles 6-10).
Illegal Access (Article 7): This article risks criminalizing essential activities in security research, particularly where researchers access systems without prior authorization to identify vulnerabilities.
Illegal Interception (Article 8): Analysis of network traffic is also a common practice in cybersecurity; this article currently risks criminalizing such analysis and should similarly be narrowed to require malicious criminal intent (mens rea).
Interference with Data (Article 9) and Interference with Computer Systems (Article 10): These articles may inadvertently criminalize acts of security research, which often involve testing the robustness of systems by simulating attacks that could be described as “interference” even though they don’t cause harm and are performed without criminal malicious intent.
All of these articles fail to include a mandatory element of criminal intent to cause harm, steal, or defraud. A requirement that the activity cause serious harm is also absent from Article 10 and optional in Article 9. These safeguards must be mandatory.
What We Told the UN Drafters of the Convention in Our Letter?
Earlier this year, EFF submitted a detailed letter to the drafters of the UN Cybercrime Convention on behalf of 124 signatories, outlining essential protections for coders.
Our recommendations included defining unauthorized access to include only those accesses that bypass security measures, and only where such security measures count as effective. The convention’s existing language harks back to cases where people were criminally prosecuted just for editing part of a URL.
We also recommended ensuring that criminalization of actions requires clear malicious or dishonest intent to harm, steal, or infect with malware. And we recommended explicitly exempting good-faith security research and investigative journalism on issues of public interest from criminal liability.
What Has Already Been Approved?
Several provisions of the UN Cybercrime Convention have been approved ad referendum. These include both complete articles and specific paragraphs, indicating varying levels of consensus among the drafters.
Which Articles Has Been Agreed in Full
The following articles have been agreed in full ad referendum, meaning the entire content of these articles has been approved:
Article 9: Interference with Electronic Data
Article 10: Interference with ICT Systems
Article 11: Misuse of Devices
Article 28(4): Search and Seizure Assistance Mandate
We are frustrated to see, for example, that Article 11 (misuse of devices) has been accepted without any modification, and so continues to threaten the development and use of cybersecurity tools. Although it criminalizes creating or obtaining these tools only for purposes of violations of other crimes defined in Articles 7-10 (covering illegal access, illegal interception, interference with electronic data, and interference with ICT systems), those other articles lack mandatory criminal intent requirements and a requirement to define “without right” as bypassing an effective security measure. Because those articles do not specifically exempt activities such as security testing, Article 11 may inadvertently criminalize security research and investigative journalism. It may punish even making or using tools for research purposes if the research, such as security testing, is considered to fall under one of the other crimes.
We are also disappointed that Article 28(4) has also been approved ad referendum. This article could disproportionately empower authorities to compel “any individual” with knowledge of computer systems to provide any “necessary information” for conducting searches and seizures of computer systems. As we have written before, this provision can be abused to force security experts, software engineers, tech employees to expose sensitive or proprietary information. It could also encourage authorities to bypass normal channels within companies and coerce individual employees—under threat of criminal prosecution—to provide assistance in subverting technical access controls such as credentials, encryption, and just-in-time approvals without their employers’ knowledge. This dangerous paragraph must be removed in favor of the general duty for custodians of information to comply with data requests to the extent of their abilities.
Which Provisions Has Been Partially Approved?
The broad prohibitions against unauthorized access and interception have already been approved ad referendum, which means:
Article 7: Illegal Access (first paragraph agreed ad referendum)
Article 8: Illegal Interception (first paragraph agreed ad referendum)
The first paragraph of each of these articles includes language requiring countries to criminalize accessing systems or data or intercepting “without right.” This means that if someone intentionally gets into a computer or network without authorization, or performs one of the other actions called out in subsequent articles, it should be considered a criminal offense in that country. The additional optional requirements, however, are crucial for protecting the work of security researchers and journalists, and are still on the negotiating table and worth fighting for.
What Has Not Been Agreed Upon Yet?
There is no agreement yet on Paragraph 2 of Article 7 on Illegal Access and Article 8 on illegal interception, which give countries the option to add specific requirements that can vary from article to article. Such safeguards could provide necessary clarifications to prevent criminalization of legal activities and ensure that laws are not misapplied to stifle research, innovation, and reporting on public interest matters. We made clear throughout this negotiation process that these conditions are a crucially important part of all domestic legislation pursuant to the convention. We’re disappointed to see that states have failed to act on any of our recommendations, including the letter we sent in February.
The final text dated May 23, 2024 of the convention is conspicuously silent on several crucial protections for security researchers:
There are no explicit exemptions for security researchers or investigative journalists who act in good faith.
The requirement for malicious intent remains optional rather than mandatory, leaving room for broad and potentially abusive interpretations.
The text does not specify that bypassing security measures should only be considered unauthorized if those measures are effective, nor make that safeguard mandatory.
How Has Similar Phrasing Caused Problems in the Past?
There is a history of overbroad interpretation under laws such as the United States’ Computer Fraud and Abuse Act, and this remains a significant concern with similarly vague language in other jurisdictions. This can also raise concerns well beyond researchers’ and journalists’ work, as when such legislation is invoked by one company to hinder a competitor’s ability to access online systems or create interoperable technologies. EFF’s paper, “Protecting Security Researchers' Rights in the Americas,” has documented numerous instances in which security researchers faced legal threats for their work:
MBTA v. Anderson (2008): The Massachusetts Bay Transit Authority (MBTA) used a cybercrime law to sue three college students who were planning to give a presentation about vulnerabilities in Boston’s subway fare system.
Canadian security researcher (2018): A 19-year-old Canadian was accused of unauthorized use of a computer service for downloading public records from a government website.
LinkedIn’s cease and desist letter to hiQ Labs, Inc. (2017): LinkedIn invoked cybercrime law against hiQ Labs for “scraping” — accessing publicly available information on LinkedIn’s website using automated tools. Questions and cases related to this topic have continued to arise, although an appeals court ultimately held that scraping public websites does not violate the CFAA.
Canadian security researcher (2014): A security researcher demonstrated a widely known vulnerability that could be used against Canadians filing their taxes. This was acknowledged by the tax authorities and resulted in a delayed tax filing deadline. Although the researcher claimed to have had only positive intentions, he was charged with a cybercrime.
Argentina’s prosecution of Joaquín Sorianello (2015): Software developer Joaquín Sorianello uncovered a vulnerability in election systems and faced criminal prosecution for demonstrating this vulnerability, even though the government concluded that he did not intend to harm the systems and did not cause any serious damage to them.
These examples highlight the chilling effect that vague legal provisions can have on the cybersecurity community, deterring valuable research and leaving critical vulnerabilities unaddressed.
Conclusion
The latest draft of the UN Cybercrime Convention represents a tremendous failure to protect coders’ rights. By ignoring essential recommendations and keeping problematic language, the convention risks stifling innovation and undermining cybersecurity. Delegates must push for urgent revisions to safeguard coders’ rightsandrights and ensure that the convention fosters, rather than hinders, the development of a secure digital environment. We are running out of time; action is needed now.