Know Your Contacts: Usable Authentication for Secure Messaging
My master's thesis redesigns contact verification in end-to-end encrypted messengers like Signal. Users preferred it, but only automation got more contacts verified.

Hi, I'm Tobi. I work on usable security and privacy. Right now I'm mostly focused on building twonly, a privacy-friendly Snapchat alternative. Here I write about what I'm working on.
My master's thesis redesigns contact verification in end-to-end encrypted messengers like Signal. Users preferred it, but only automation got more contacts verified.
Users switching from Snapchat to twonly want to take their Memories with them. I designed a protocol that lets a browser encrypt and upload thousands of memories into a user's end-to-end encrypted twonly Memories, without the browser ever receiving a key that could read, modify or delete anything else.
I designed a passwordless recovery protocol using trusted friends and a server-side second factor for my privacy-focused messaging app twonly. To ensure its security before release, I am launching a 50€ bug bounty challenge for anyone who can find a vulnerability in the protocol.
I wrote a seminar paper on binary rewriting, which allows the modification of compiled binary files without source code, enabling use cases such as fuzzing or hardening proprietary binary files.
This blog post introduces my new generic approach to easily create a fast and easy-to use protocol fuzzer for custom targets.
twonly is a European open-source alternative to Snapchat, written in Flutter. It uses the Signal protocol to encrypt all messages end-to-end.
This winter semester I took part in the TU Darmstadt Hacker Contest, where we had an exercise in which we had to find security vulnerabilities in open source repositories.
In my term paper about the "Internet of Vulnerable Things" I wanted to find a memory-related vulnerability in a binary running on the TL-WR902AC but was not successful. This time I use fuzzing.
To learn Rust and cryptography at the same time I implemented the TLSv1.3 from scratch. The implementation includes all cryptographic operations like elliptic curves or AES.
In one of my term papers I had to write about the topic "Internet of Vulnerable Things". So I bought a cheap router and took a closer look. As expected, the security was not really good and I was able to find a security vulnerability with a CVE score of 8.8 in no time.