 Hello everyone, it's a great honor to share our work on Glitching attacks of SHA-3 with you at Azure Crypt 2022. So in this small talk I will give a brief introduction of the previous works and our progress. Let's first have a look at the SHA-3 hash family. It consists of six instances and each of which is based on the sponge construction with K checkup permutations as the underlying permutation. In the previous works, at most five runs are covered and in this work we present the first six-round Glitching attacks on SHA-3. The previous works take a three-step framework that is first generate the gliding trail and then construct connectors and add the last exhaustive research for glitching pairs. The limitations are from two aspects. For gliding trail it's difficult to find better trails cause it's highly dependent on the general trail search which is a difficult problem itself and for the connected construction due to the quick consumption of the degree of freedom it's impossible to extend one more round. In this work we propose a set-best trail search toolkit that generates better gliding trail and the connecting trail and based on which we command six-round glitching attacks. Similar to other two best automatic trail search approaches, the set implementation is quite generic. We first translate the problem to languages that the software can understand and then we solve it. In this work we choose script to mini-set as our software and describe the propagation as well as the objective function with the same asset that the software can understand. With the set-best toolkit we implement the truncated differential trail search, the gliding trail search and the connecting trail search with better results and good efficiency. So with the set-best toolkit the attack strategy of the collision attack is direct. We first generate a four-round gliding trail and then we determine the two or three-round connecting trail and in the connector construction we only leave the equations on messages and the solution space are a subset of message pairs that will follow or partially follow the connecting trail. Then in the exhaustive search phase we will enumerate the message pairs until a gliding pair is obtained. So in total we propose the first six-round glitching attacks on char theory instances. That includes six-round collision attacks on shake one to eight in both classical and quantum settings and the two quantum collision attacks on char theory to five sticks and the two to four. Okay that's all for our work. Thanks for your listening.