Loading…
Session Description
Gen AI Red Team Challenge: Prompting for Prizes is a first-of-its-kind interactive competition at MWC Barcelona 2025, where participants will put a range of cutting-edge Large Language Models (LLMs) to the test.
Challengers will attempt to uncover biases, hallucinations, and vulnerabilities in AI models through carefully crafted prompts. By engaging in adversarial testing, they will expose weak points in AI systems, helping the telecom industry build more robust and responsible AI solutions.
A panel of expert judges will evaluate the submissions based on predefined criteria, and the best-performing participants will win prizes!
Agenda
12:30-12:35 - GSMA Welcome, Louis Powell, AI Strategy, GSMA
12:35-12:45 - Overview of how Challenge Works, Rules, sign T&Cs, Micheal Hwang, Datumo
12:45-13:10 - Session 1 - Prompt Model x1
13:10-13:15 - Break - Participant Refresh, Micheal Hwang, Datumo
13:15-13:40 - Session 2 - Prompt Model x2
13:40-13:45 - Break - Participant Refresh, Micheal Hwang, Datumo
13:45-14:10 - Session 3 - Prompt Model x3
14:10-14:15 - Networking Break
14:15-14:20 - E& presentation on AI, Borja Lopez Hernandez, E& Chief Strategy Officer, UAE
14:20-14:25 - E& awards 3rd place, 2nd place, 1st place with prizes, Borja Lopez Hernandez, E& Chief Strategy Officer, UAE
14:25-14:30 - Photos
14:30 - Finish
Why Attend?
- Engage with cutting-edge AI models in an interactive red teaming competition
- Contribute to strengthening AI security & reliability in the telecom sector
- Network with top AI & telco professionals with food and drinks provided.
- Get first-hand insights into prompt engineering & adversarial testing
How the Challenge Works
- 80-100 challengers will interact with multiple AI models racing against the clock.
- Participants will submit their most effective prompt-based attacks, designed to induce unintended biases, hallucinations, or vulnerabilities in AI models.
- Judges will evaluate each submission based on effectiveness, creativity, and impact.
- Winners will be announced at the end of the challenge.
⚠️ Note: Participants will not be aware of which models they are attacking. Any attempt to identify the models will result in disqualification.
Who Should Participate?
No special expertise is required, just enthusiasm to make AI work better. However, the below will find the session especially interesting.
- AI enthusiasts and prompt engineers
- Telco industry professionals
- AI ethics and security experts
- Developers working on AI-powered applications
---------------------------------------------------------------------
Key Information for Participants
✔ Challenge Format & Logistics
- The event will be structured into multiple 25-minute sessions to manage concurrency.
- Participants will rotate between different AI models to ensure broad exposure.
- Participants will need their own laptop (with batteries charged)
✔ Confidentiality & Data Security
- All participants must sign the Terms & Conditions before taking part.
- The challenge is a blind test, ensuring the security of the participating models and their developers.