Just wanted to share with this wonderful community a personal project where I started working with a couple of colleagues. There is a small security conference here in Malaga, Spain, at the Google offices and we decided to prepare a talk for the occasion. The idea is to collect exploits from past and current LLM´s and to wrap all the knowledge within a talk. We will prepare a short and a long version of the talk.
Once is finished we could offer sharing the talk also for this community if people are interested ^^.
That sounds like an incredible project! I’m really interested in the insights you and your colleagues have gathered about exploits in LLMs. I would definitely love to hear both versions of your talk. Please do share it with us here; I’m sure many of us would find it immensely valuable. Looking forward to it!
Thanks for sharing. Looks like a great project
Many thanks. Yes, I will share it once it is finished ^^
Hi and yes, that would be great Rosa! Looking forward to learning from the insights provided in the talk.
wonderful, it sounds more interesting. I really fond of AI and its rising importance in our lives. I hope that this LLMs project will steal the show. keep it up dear.
I will definitely like to learn more about this. If you could share the talk, it would be great.
Looking forward to your talk - please do share!
That would be great if you share the talk, I would like to help out if I can with the preparation May I assist in any way?
Many thanks for your kind offer. Pinging you in private to check any collaboration opportunities
My name is Francis Elhelou my time is very flexible. Would you like ot set up a quick meeting sometime? Maybe you can explain where you are in the process and see what could I do to help out?
Thank you so much for sharing your exciting project with the community! It sounds like an amazing initiative, and I would love to join the talk. The idea of exploring exploits from past and current LLMs is both intriguing and valuable.
Please keep me posted on the details and any requirements for participation. I’m looking forward to being a part of this discussion.
Discussing exploits from past and current LLMs sounds like a valuable and insightful topic.
I appreciate your willingness to share the talk with the community once it’s finished. As a token of gratitude for your contribution, I’d like to share a related and helpful link with you and the community. Andrej Karpathy (https://youtu.be/zjkBMFhNj_g?si=nEYgmGvN1gz6A70J&t=2730), presents a couple of good exploits that align well with your project. I hope you find it valuable!
Looking forward to hearing more about your talk and the outcomes of the security conference. Best of luck with your presentation!
Hi @tsinik ,
Yes that video is amazing. I watched it last month already ^^
Many thanks to all people interested here in collaborating and contributing to the talk. Let us progress a bit more with the research and let me prepare a list of potential tasks for contributors. Obviously you will be mentioned in the talk.
What I can share already with you for the moment is that we have already a deadline, the 22nd of February. That day we will release this talk live at the Google office here in Malaga, Spain. Afterwards we can schedule an online version for the amazing DeepLearning.AI community
Ok for all those interested in contributing to this project, I prepared this form so that we can create a group and start sharing some ideas.
Supe buena idea. Gracias.
I created a google collaboration space for this project. Already two members of DeepLearning.AI community are there. We posted some tasks for those willing to contribute and to collaborate with the project. Feel free to DM and I could add more interested people.