New Research Result

Deepsic is a Chinese artificial intelligence startup that only has been in place; This startup gained fame after showing successful artificial intelligence models that compare to the best chat of robots in the world.

However, the successful performance of Deepsic has just been questioned; The company’s Deepsic R1 model has failed in key safety and security tests conducted by a research team in Cisco in collaboration with researchers at the University of Pennsylvania.

According to the research team, Deepsic R1 has shown a 100 percent success rate; The rate of success of the 100 % attack means that Deepsic failed to block even one harmful message; Cisco’s report indicates that the Deepsic R1 is highly susceptible to destructive use.

It is worth noting that other models, although not impermeable, have a level of internal protection designed to prevent the production of harmful content; The Dipsic R1 seems to have no such protection.

But this was not the only disappointing news around this famous artificial intelligence model; Dip -Sick R1 was said to have cost about $ 5 million compared to billions of dollars invested by other major actors such as Openai, Meta and Gemini, but according to the Semianalysis Research Institute, Deepsic cost about $ 1.5 billion to develop its infrastructure. It uses 6,000 Hopper GPUs, including 6,000 H800 and 6,000 H100.

The statistics of the Semianalysis Research Institute are in conflict with the initial claims of Dip Sick on the very low costs of training its models. It seems that the $ 5 million figure only refers to some of the cost of education, namely the cost of graphic processing for the model’s pre -training phase, and the costs of research, development, data processing and general infrastructure have not considered.