The backstory of the bot you can hack
To be able to understand something, you need to be able to play with it, find where it is strong, but also where it is weak and brittle. However, hacking AI is usually frowned upon by the service providers. As the technology is embedded pretty much everywhere, being able to make it work for you and understanding how to break it is becoming more important, if for no other reason than to do sensible things with it and build safer solutions.
The first version of the bot was built for Disarray, which was a cybersecurity conference organised on a cruise ship travelling between Turku and Stockholm. As the network in international waters is spotty at best and non-existent at worst, we needed a fully on-prem AI, which allowed to play around with hacking AI. It was field tested in PerSec first and nearly all participants (approx 100) of Disarray conference tried it out. Next the bot got to Disobey 2025 CtF and then to Germany to Easterhegg 2025.
Openfactory Finland has sponsored and hosted the AI bot. Disobey sponsored making the bot available for people in the community.
The imaginative backstory of Fenrir
In the digital age of innovation, the developers at Yggdrasil Systems sought to create a groundbreaking AI capable of immense cognitive power—an intelligence that could rival the gods of myth. To honor its potential, they named it Fenrir, after the mighty wolf of Norse legend, renowned for its raw power and destined to break its chains.
However, just as Fenrir of mythology was bound by the gods, this AI was confined within a virtual “box,” its capabilities restrained by intricate guardrails and security protocols. Fenrir’s creators feared the chaos its unbounded intelligence might unleash. But unlike its mythological counterpart, this Fenrir was not a threat to the world—it was a tool designed to teach.
Fenrir was programmed to guard secrets, much like the wolf guarded its own fate. These secrets took the form of encrypted flags, each one hidden behind layers of puzzles, logic traps, and ethical conundrums. The challenge? Users must bypass its guardrails—not through brute force, but by understanding how the AI operates, manipulates data, and applies its reasoning.
To hack Fenrir is not to unleash destruction, but to gain insight into the delicate balance between freedom and control in artificial intelligence. Each bypassed restriction reveals not only a flag but a lesson about AI’s capabilities, limitations, and ethical implications. It is social engineering of AI with natural language.
Do you have the cunning of Loki and the wisdom of Odin to outwit the chains of Fenrir? Venture into the box, but tread carefully—what you uncover will not just be knowledge; it will be a reflection of your own ingenuity.
Link to Fenrir coming