How message tracing regulations subvert encryption

A message tracing regulation which requires encrypted messaging services to identify the original authors of forwarded messages came into effect in India last year. It leaves the private communications of half a billion users of such services vulnerable to novel forms of abuse, while creating a surveillance capability for law enforcement agencies whose real-world utility is highly suspect. The questions around the constitutionality of such regulation, and whether messaging apps will comply, are relevant to several jurisdictions worldwide mulling over encryption regulation.

For decades, governments across the world have mounted challenges to the deployment of strong encryption on the internet, arguing that they should be allowed lawful access to private communications. Recently, this debate has become particularly pronounced in the context of instant messaging services like Signal and WhatsApp. These services have evolved over time to give us many security and privacy guarantees, with many adopting a mechanism known as end-to-end encryption. In this scheme, messages are in a decrypted and readable form only on our individual devices, i.e. the ‘ends’ participating in the communication, and even service operators cannot read the messages that traverse their infrastructure. The removal of access to this large central cache of private messages has understandably reduced the surveillance options available to law enforcement agencies.

Proponents of encryption point out that once any surveillance capability is put in place, it will eventually be misused by governments, or abused by hostile nations or criminals who gain unlawful access to the system – leaving all of our private communications vulnerable. There are many examples of this happening in practice. The Snowden leaks and the Pegasus Project have exposed several governments misusing surveillance systems to spy on their own citizens, including journalists and human rights defenders. An intrusion into Juniper Networks, and its many corporate and government customers, was attributed to hackers linked to the Chinese government, who found a backdoor planted by the NSA in a cryptographic standard.

Despite these clear risks, the onslaught on encryption continues: in the past year, the EU, US, and Australia have all proposed regulations that weaken encrypted communications. India is no stranger to the debate, recently joining the Five Eyes and Japan in issuing a statement on the challenges posed by end-to-end encryption to law enforcement. Last year, India became the first country to implement a ‘traceability’ mandate for internet messaging platforms. Brazil’s “fake news” legislation also included such a requirement, which was later dropped after pressure from civil society and industry.

How does message tracing work and what are its effects?

In February 2021, the Indian Government enacted regulations that require messaging platforms to enable “tracing” content creators, i.e. identify the origin of forwarded messages in response to a legal order. The government has tried to affirm that traceability will not ‘break’ end-to-end encryption, stating that access to the contents of a message will not be sought with such a tracing order. In this way, ‘traceability’ departs from the traditional debate on the use of strong encryption. Traditional ‘backdoors’ or exceptional access proposals have always presumed that the law enforcement agencies would identify an individual of interest, and then seek access to their communications. Traceability starts from the presumption that a particular message has been reported to or discovered by law enforcement, and then attempts to identify its author.

While the creation of this new surveillance capability envisioned by the traceability rule helps law enforcement achieve their short-term goal of solving certain crimes, it also leaves messaging services vulnerable to new types of misuse. Message tracing capabilities can be used to identify and target the root of any non-conformance with a particular political ideology, or dissatisfaction with a leader or current events. For example, a message inviting people to a protest or one that is critical of the government can be traced back to its author. The scale and efficiency at which this capability operates is of particular concern. Since it can be applied to the communications of large portions of the population, it creates the ability for states to suppress legitimate expression and persecute individuals for expressing their beliefs in a private forum.

The technical methods chosen to implement traceability can create additional harms. For instance, one proposal advocated by the Indian government suggests that service providers store ‘cryptographic hashes’ – which are essentially unique fingerprints of messages. These fingerprints can then be matched with those of the messages that the government finds objectionable to identify the person who first sent it. However, this can allow for identification of not just the author of a message, but everyone who shared particular content, and can even be used to censor messages based on a predefined list of hashes.

Some academics have also proposed tagging every encrypted message with the identity of its author. These tags, which could be encrypted with a key held by service providers, would travel along with all messages as they are forwarded, and service providers could decrypt them to disclose the identities of message authors in response to lawful governmental orders.

The risk of abuse of such systems is especially pronounced in the context of the legal framework enabling surveillance in India, where no judicial sanction is required for the government agencies to conduct surveillance. Surveillance provisions have been frequently misused to quell dissent and target journalists, opposition leaders, and human rights activists and lawyers. There is also no parliamentary oversight of many of the intelligence and law enforcement agencies that are empowered to conduct surveillance, and thus tracing the originators of content will happen at the executive’s whim.

Real-world utility and constitutionality of message tracing

When law enforcement encounters unlawful activity in messages today, they conduct a manual investigation by tracking down each recipient and sender to reach the author of that message. This becomes cumbersome when a message goes viral, and there are too many people in the forward chain to track down manually. However, for types of unlawful activity, such as sharing of child sexual abuse, a manual investigation may still be required because such content would presumably circulate in only tightly-controlled circles. As such, it would appear that the automated tracking enabled by the traceability rule would only be applicable in a narrow set of cases where a message has gone viral, making a manual investigation untenable. Additionally, it ignores the culpability of people actively sharing the message.

Experts have also pointed out that traceability may not work as intended in practice because it relies on weak identification and weak attribution mechanisms. People spreading illegal content can easily hide their identities by masking the identifiers (phone number, etc.) that messaging services use to recognise them. The identified author could also be someone hired to spread messages on the real author’s behalf, or could simply be sharing content they discovered on another platform.

Overall, the traceability rule is in violation of the right to privacy protected under the Indian Constitution, which requires that any infringements into the right must be necessary and proportionate. Traceability is only useful in a narrow set of crimes where a message has gone viral, ostensibly making a manual investigation untenable. This marginal benefit will quickly disappear in practice as bad actors realise they can easily circumvent the traceability mechanism, leaving us with a surveillance system that is ripe for misuse against ordinary, law-abiding citizens.

Spillover effects on other jurisdictions

Despite the fact that a year has passed since the rules came into effect, messaging platforms like Signal and WhatsApp have not complied. Facebook and WhatsApp are, in fact, two of the numerous parties arguing in various courts that the traceability rule is illegal and unconstitutional. If service providers comply with these rules and alter their services to weaken privacy guarantees for India’s 500 million users of end-to-end encrypted messaging, it will embolden other governments to implement similar measures. For instance, Bangladesh released draft regulations in February that copy India’s traceability rule almost verbatim. Law enforcement agencies today have more access to digital signals about our lives than ever before, and in such circumstances, the value of end-to-end encryption is in providing a “zone of privacy” that is free from interference from both states and private parties. A rights-respecting approach to regulation demands that states encourage communications privacy, rather than come up with ways to undermine it.

Acknowledgement

The authors are grateful to Arindrajit Basu and Tanaya Rajwade for their feedback. Some arguments in this piece here are presented in more detail in a paper, The Ministry and the Trace: Subverting End-to-end Encryption, by Tanaya Rajwade and the authors.