If the promised revolution of the business world by artificial intelligence takes place, it will not happen all at once. Four in ten CTOs believe their company is not ready for the advent of new automation systems like OpenAI’s ChatGPT.
This statistic emerges from a survey conducted at the end of March on behalf of the Californian giant of data centers and Equinix infrastructures with 2,900 IT managers in around thirty countries on four continents.
The problem for these companies is twofold. Generative AIs like ChatGPT promise to revolutionize the way we create value from big data and digital technologies. In its own study published in early spring, Goldman Sachs predicted that these AIs could add 7% in value to global GDP. Companies that aren’t ready to adopt them risk missing the boat, warns Equinix.
It’s very difficult to prevent all incidents — but you have to limit the damage and then adjust
Second, and perhaps more seriously, companies whose computer systems are not adequately adapted to the adoption of these new tools could become prime targets for hackers.
In the Equinix survey, 85% of IT managers hope to take advantage of advances in AI to improve business performance. Those among them who rush to adopt the latest technologies without first securing their own digital assets could pay the price, says Kaladhar Voruganti, advanced technology expert for Equinix. “For example, a company that decides to create its own generative AI application might want to have AI models located outside of its facilities process sensitive data distributed across various internal and external centers. »
These numerous data exchanges represent so many potential breaches that could be exploited by Internet users with malicious intentions, we fear at Equinix.
The cat and the Mouse
Equinix President for the Americas, Tara Risser, was at the Conference of Montreal on Monday to discuss with other IT experts the security risks associated with the emergence of new artificial intelligence applications.
At least M.me As a result, Risser does not fear the advent of this new generation of AI the digital equivalent of a nuclear war, which several well-known AI experts seem to fear, including Montreal researcher Yoshua Bengio.
This does not mean that there are no security risks associated with AI. Alongside his colleague from Equinix on stage during the conference, Jean-François Leduc, vice-president and chief technology officer for the Montreal-based multinational security company GardaWorld, compared the evolution of technologies to a game of cat and mouse. .
“I don’t see in AI a threat of the end of the world,” he said in an interview with the Duty. “It’s a new technology that also makes it possible to create new security tools that will better protect us. Obviously, the risk of AI being misused is there, but there are also people doing good things with it. »
Regardless, warns Jean-François Leduc, companies that intend to adopt AI tools must be prepared for possible security breaches. “There are always risks,” he says. “It is very difficult to prevent all incidents — but you have to limit the damage and then adjust. »
Automated Cyber Threats
Hence the importance, both for companies that plan to bet on AI and for those that are not interested in it, to at least offer clear information to their employees on the right ways to to use, or not, and what to do if there is a computer security incident at work.
Because there are cyberpirates who use them, these technologies. They help them refine their practices. ChatGPT is able to understand and then generate text in about fifty languages. Writing an impeccable fake email in several languages, including French, which will be used to lure people in business to extract confidential information or money from them, suddenly requires very little effort, notes the head of GardaWorld technologies.
“Executives who don’t believe they could be targeted forget that any company that has information is a potential target — even an SME with 10 employees. Attackers no longer target specific organizations, they seek out the most vulnerable. Their tools go around the Web to find organizations that have weaknesses and then target them, going to retrieve sensitive information from their website or social networks to personalize their attacks a little more.
All of this is automated.
“A ransom of $200 obtained in three minutes, if the tool does everything automatically, it pays! It may not be the end of the world, but it is an issue that companies should take seriously, concludes the computer security expert.