At FOSDEM 2019, Belgium, Maximilian Blochberger talked about preventing cryptographic pitfalls by avoiding mistakes while integrating cryptographic mechanisms correctly. Blochberger is a research associate at the University of Hamburg. FOSDEM is a free and open event for software developers with thousands of attendees, this year’s event took place on second and third February.
The goal of this talk is to raise awareness of cryptographic misuse. Preventing pitfalls in cryptography is not about cryptographic protocols but about designing better APIs.
Consider a scenario where a developer that values privacy intends to add encryption. This is about integrating cryptographic mechanisms into your application. Blochberger uses a mobile application as an example but the principles are no specific to mobile applications. A simple task is presented—to encrypt a string which is actually difficult. A software developer who doesn’t have any cryptographic or even security background would search it online.
They will then copy paste a common answer snippet available on StackOverflow. Even though it had warnings of not being secure, but had upvotes and probably worked for some people. Readily available code like that has words like “AES” or “DES” and the software developer may not know much about those encryption algorithms. Using the default algorithms listed in such template code, and using the same keys is not secure. Also, the encryption itself is not CPA (chosen-plaintext attack) secure, the key derivation can be unauthenticated, among other things.
98% of security-related snippets are insecure according to many papers. It’s hard to get encryption right. The vulnerability is high especially if the code is copied from the internet. Implementing cryptographic mechanisms should be done by cryptographic engineers who have expertise in the field. The software developer does not need to develop or even know about the details of the implementation.
Doing compiler checks instead of runtime checks is better since you don’t have to wait for something to go wrong before identifying the problem.
Cryptography is harder than it actually looks. Many things can and do go wrong exposing encrypted data due to incorrect choices or inadequate measures. He demonstrates an iOS and macOS example using Tafelsalz.
For more details with the demonstration of code, you can watch the video.
Read next
Introducing CT-Wasm, a type-driven extension to WebAssembly for secure, in-browser cryptography
Tink 1.2.0: Google’s new multi-language, cross platform, cryptographic library to secure data