Pavel Korostelev, Head of Product Promotion at Code Security LLC,

[tanjimajuha20]

A more serious problem, as Nikolai Domukhovsky noted, is the problem of the quality of datasets for training AI. He complained that only large players in the information security market or companies affiliated with large digital platforms can collect high-quality datasets. The representative of the UCSB called the datasets that can be found in the public domain nothing less than "garbage".

According to Ivan Rybnikov, integrating tunisia whatsapp resource AI with existing information security systems can be a complex task that requires significant resources and time: "False positives, both false positives and false negatives, can reduce trust in AI systems. In addition, the effective use of AI requires large volumes of high-quality data, which is sometimes difficult to provide. It is also necessary to take into account the difficulties in adapting AI to constantly changing types of cyberattacks and threats."

called the task, the solution of which is greatly simplified by generative AI, "translation" of software source texts from one programming language to another. He noted that in the conditions of a shortage of developers, this arises more and more often due to the fact that the existing developers simply do not know the language in which a particular application was originally written.

The downside, as Pavel Korostelev warned, is the risk of data leaks, which some corporations have already encountered, in particular, Samsung. As a result, in his opinion, the development of large language models has given rise to a discussion about technological sovereignty in the EU and India against the backdrop of possible leaks of technological secrets to the US or China. At the same time, Pavel Korostelev expressed concern that AI will increase the workload of information security specialists, since reducing routine tasks will free them up to solve problems that they simply did not have time for before.