OpenAI made AI applications available to everyone in November 2022 when it launched its ChatGPT Program. The chatbot made AI accessible to everyone, not just computer scientists and data scientists.
But not everyone is interested in using these tools for benign reasons. In Hong Kong, a finance employee transferred over $25 million to scammers who used deepfake to pose as the chief financial officer of the company on a video chat.
Maria Milosavljevic said, “These things keep me awake,” during a Fortune online conversation held on Wednesday. “Unfortunately, AI, which is extremely useful and powerful, is accessible to both our enemies and friends.”
The event, held in partnership with Accenture explored the interaction between new generative AI and cybersecurity.
Scott Wilkie, global leader for emerging technology security at Accenture, said that there is a “widening of the breadth of attacks across the boards.” The consulting firm has seen a “doubling” in ransomware attacks and a 1,000% rise in phishing over the last 12 months.
“Generative AI, and new large language model models, enable more sophisticated attacks at a greater volume,” said he.
Calvin Ng, the director of the Cyber Security Agency of Singapore’s cybersecurity program center, acknowledged the increase in the frequency and severity of attacks. He explained that proper risk assessment and management is necessary.
Ng explained: “You can easily create a phishing e-mail, you can automate malware.” “You don’t have to be a cybersecurity expert to create malware with ChatGPT. Things are becoming simpler; doing evil has become easier today.
Ng warned against the possibility of “data-poisoning,” in which an adversary targets the AI mode’s training set. He said that organizations should consider the implications of AI that can “trawl through information from all over the world and produce information without consulting anyone” and implement guardrails to avoid data poisoning.
Microsoft announced on Wednesday that it had found a way to jailbreak an AI model that generates content related to explosives. drugs and politics.
Panelists noted that AI can be used to help cybersecurity teams, not hinder them.
“We receive more than 10 billion data event notifications every day.” Milosavljevic explained that because humans can’t look at everything, 35% of incident response is already automated.
Cybersecurity is a team sport
As cyber attacks continue to grow, cybersecurity is no longer a concern of IT departments alone. The panelists on Wednesday noted that cybersecurity was not just a company-wide initiative but a collaborative effort that included multiple parties, including national governments.
Wilkie said, “We have always believed that cybersecurity is a team sport.” “I have never seen collaboration be more effective or well-intentioned than it has been in the last five years.”
, for example, is an international pact that includes 40 countries to not give in to ransomware attacks. Members also agree to collaborate on research projects that will build resilience.
Geopolitics and national security are also reasons why governments care. Ng said that while technology can help a nation grow its economy, this must be done in a “balanced” environment.
Jennifer Tiang, regional director of cyber practice at Willis Towers Watson for Asia, said that companies, especially those outside of industries heavily regulated, need to “get their basic cyber hygiene right”. She compared it to home security. Even the most sophisticated cameras are useless if the owners don’t lock their doors.
She said, “The risks they face are very sophisticated.” “They must get the basics right and invest in areas where there may have been minimal investment.”
Fortune brings Brainstorm AI for the first-time to Asia, bringing together executives, founders and venture investors from around the globe in Singapore, from July 30 to August 31. Register here!