Youtuber and programmer Howdy Ho has developed a neural network for Counter-Strike 2, which can shoot the enemy itself, imitating human actions, and VAC will not detect it in any way. According to the youtuber, its creation required titanic labor, a lot of calculations and 15-20 thousand game screenshots with player models from CS2 for its training. Howdy Ho used advanced machine learning techniques to train his neural network, which can quickly and accurately detect enemies on the screen and automatically fire shots.

The main reason why the antichit does not see the neural network working is because it is powered by OBS. The neural network captures the screen image, analyzes it and makes decisions based on the visible content without changing the game files or interfering with the game code. By all checks, it looks like you're streaming or recording a Counter-Strike 2 video, making it invisible to anti-cheat systems.

READ ALSO: Lynn Vision signs flying after successful test performance

Haudi Ho also stated that if we continue working on this neural network, it will be able to automatically walk and aim like real players, not differentiating in behavior from the best professional gamers. In his video, he explained the technical details of the development and left the source code of the project for those who have free time and desire to continue the experiment.

The future of such technologies raises many questions. How will anti-hit systems adapt to new methods of circumvention? Will game developers be able to find ways to protect themselves from such sophisticated tools?

Source: YouTube and Github