AI security 2024: Key insights for staying ahead of threats - helpnetsecurity.com/2024/08/08/koj… - Robust Intelligence Kojin Oshiba #ArtificialIntelligence #GenAI #RiskManagement #Regulation #CyberSecurity #netsec #security #InfoSecurity #CISO #ITsecurity #CyberSecurityNews #SecurityNews

Very strong password.

follow TMLR Papers with Videos for awesome video recordings associated with Accepted papers at TMLR !

A breakthrough just dropped. arxiv.org/pdf/2408.03583

In the beginning, ML was RL. And here is a very nice book on the mathematical foundation of reinforcement learning. github.com/MathFoundation…

How did your NeurIPS Conference rebuttal go?

I just bought the most expensive cappuccino in my life from the Copenhagen airport: $10. Kasper Green Larsen made false promises.

Our research on bypassing safe structured output of OpenAI openai.com/index/introduc…

Denmark, what a beauty.

Thrilled to announce Cisco's intent to acquire Robust Intelligence 🎉 From a windowless office in Harvard Square to transforming AI security globally—excited to unlock the next chapter with one of the biggest security companies in the world!!! More here: blogs.cisco.com/news/fortifyin…

Super proud of the whole team at Robust Intelligence and of joining force with Cisco. This is a great day for the future of our company and the future of secure AI.

Unveiling the Statistical Foundations of Chain-of-Thought Prompting Methods ift.tt/o8wZl0j

🚀 Exciting News! 🚀 Our paper "Learning Integral Operators via Neural Integral Equations" is published in Nature Machine Intelligence! 🤖🧠💡 nature.com/articles/s4225… 🧵 Let me share some highlights in this thread...👇

There is something mysterious about how LLMs cheat.

Unpopular opinion: If you make enough assumptions, you can prove anything (so far nothing contraversial). This is exactly what is happening in many theoretical neural network papers.

Great feature in the NYU Data Science blog on "LiveBench"—a new method for challenging lanuage models from Yann LeCun (Yann LeCun), Micah Goldblum (Micah Goldblum), Chinmay Hegde (Chinmay Hegde), Ravid Shwartz-Ziv and others... nyudatascience.medium.com/livebench-chal…

Hey collective Twitter AI hivemind: which Supervised FineTuning library should we use for our research? We are exploring libraries for SFT, e.g. TRL, Axolotl, Torchtune, or other options?

Shoot, like previous years I was 101 on the list of most influential people in AI TIME. I lost it to Scarlett Johansson.