The report published on Wednes Google revealed that hackers from numerous countries, especially China, Iran, and North Korea, use Chetbot Gemini that supports artificial intelligence to strengthen Cyber attacks on targets in the United States.
The company has revealed – at least so far – that access to publicly available to large language models (LLM) did cyber attackers more efficient, but did not significantly changed the type of attacks they usually perform.
LLM are models and who are trained, using huge amounts of previously generated content, to identify patterns in human languages. Among other things, it makes them skillful in the production of high-functional computer programs without errors.
“Instead of enabling a disturbing change, Generative AI allows actors to move to move faster and with a higher amount,” the report is stated.
Generative AI offered some advantages for low qualified and highly qualified hackers, the report says, writes Voa.
“However, it is unlikely that the current LLM will enable the penetrating abilities of the actors threats. We notice that AI landscape is in constant flow, with new AI models and agent systems that occur daily. How does this evolution take place, Google Threat Intelligence Group predicts that the landscape threat will develop a step how the actors of threats adopt new AI technologies in their operations. “
Google’s findings seem to agree with the previous survey published by another large American and Microsoft players, who have discovered a similar failure to achieve new offensive strategies for cyber attacks using public generative AI models.
The report is clarified that Google works on interfering with the activity of actors threats when identified by them.
“Ai, so far, did not change the game for offensive actors,” said America Adam Segal, director of the program of digital and cybernetic space in the International Relations Council. “It speeds up some things. The foreign actors gives a better ability to create phishing emails and find some code. But did that dramatically changed the game? No. “
It is unclear whether this could change in the future, Segal said. It is also not clear whether further development and technology will more likely use people who build a defense from cyber attacks or actors threaten to beat them.
“Historically, the defense is difficult, and technology did not solve that problem,” Segal said. “I doubt that even and that won’t do it. But we don’t know yet. “
Caleb Withers, a research associate in the Center for New American Security, agreed that they are likely to reach a kind of weapons race, as offensive and defensive applications are generative and evolved into cyber security. However, it is probably likely to balance each other in a great extent, he said.
“The default assumption should be that in the absence of certain trends that we have not yet seen, these tools should be as much useful to defenders as many offenders,” he said. “Everything that increases productivity, in general, applies equally, even when it comes to things such as the disclosure of vulnerability. If an attacker can use something to find vulnerability in the software, then the tool is a useful defense counsel to try and patch them. “
The report parses the types of threats that he spotted using Gemini in two basic categories.
The actors of advanced lasting threat (APT) refer to “Hacker activity supported by the Government, including cyber espionage and destructive attacks on the computer network”. In contrast, threats to information operations (IO) “are trying to influence an online audience in a deceptive, coordinated manner. Examples include Sock Puppet Accounts [lažni profili koji skrivaju identitet korisnika] and comment brigading [organizirani internetski napadi usmjereni na promjenu percepcije popularnosti na internetu].
The report was found that Iran hackers were the largest users of gemini in both categories of threats. Apt actors in Iran’s threats used a wide range of tasks, including collection of information about individuals and organizations, research objectives and their vulnerabilities, translation language and content creation for future online campaigns.
Google was accompanied by more than 20 APT actors supported by the Chinese Government using Gemini “to enable the scooting goals, for writing and developing and explaining the technical concepts and tried to provide a deeper access to the network after initial breakthrough.”
The APT-and supported by the North Korean state used Gemini for many of the same tasks as Iran and China, but also seemed to try to take advantage of the service in their efforts to accommodate the “secret companies” in order to facilitate theft of intellectual property.
Iran was also the largest user of Gemini when it comes to threatening information operations, making 75% of the detected use, reported Google. Iran hackers used the service to create and manipulate the content aimed at affecting public opinion and adapse that content to a different audience.
Chinese and the actors primarily used the service for research purposes, considering issues “from strategic interest in the Chinese government.”
Unlike the APT sector, where their presence was minimal, Russian hackers were more common when it comes to using gemini related to IO, but not only for creating content but also to collect information on how to create and use online AI chat bots .
Also on Wednesday, Kent Walker, the President of Global Affairs for Google and his home company, has used the company’s blog to notice potential hazards representing all the more sophisticated AI models, and invited the industry and the Federal Government “to work together on support for our national and economic security. “
“America keeps leadership in the race AI – but our advantage may not take,” Walker wrote.
Walker claims that the United States needs to keep their narrow advantage in the development of technology used to build the most advanced tools for artificial intelligence. In addition, the Government must simplify procurement rules to “enable the adoption and clouds and other technologies that change the game” by American military and intelligence agencies, and to establish public-private partnerships for Cyber Defense.
(Vijesti.ba)
