Quotes from Ryan Kalember, Executive Vice President, Cybersecurity Strategy at Proofpoint:
Phishing emails written by AI, dangerous?
"As with many new technologies, cybercriminals are quickly looking for ways to take advantage of them. The introduction of ChatGPT has raised concerns about phishing attacks. But at this stage, and based on current capabilities, concerns are exaggerated for a few reasons. Many social engineering emails are not designed to be "perfect" - they are deliberately poorly written to find people more likely to open them. That is also only part of the attack. Headers, senders, attachments and URLs are among the many other elements analyzed by robust detection technologies. Even when better formatted emails would provide a significant advantage, as in many situations where business emails are compromised, there is much more information that the threat actor needs access to. They need to know who is paying what to whom and when, and they probably already have access to that through other means. They don't necessarily need ChatGPT if they already have access to the victim's inbox and can just copy an old email."
AI Coded Malware
"Malware written by ChatGPT is no better at what malware should do, which is to evade EDRs and infect a machine more effectively than other toolkits already out there. A hacker needs something that works over and over again, and there is no evidence that ChatGPT is good at that. Hackers also need to constantly rotate more subordinate things, which ChatGPT cannot do. Attackers must rotate their infrastructure, register domains and move items because their attack attempts are constantly being picked up and destroyed by cybersecurity experts such as Proofpoint. Malware operators must also not only distribute their malware, but also sell access to it. ChatGPT makes no contribution to automating important parts of it. It is therefore unlikely to make an impact until it exceeds the capabilities of the tools that all already exist, which have been developed for this purpose for many years and do different things than a chatbot. That's not to say that it won't get as good; it's clearly moving very fast with the data set it's trained on, but it's not that far along today."
Russian hackers and ChatGPT
"Even if Russian cybercriminals try to gain access, which is fairly easy because it is free, the importance is questionable. It certainly does not replace all the tools and infrastructure developed over decades to write Windows malware. Eventually, attackers may use ChatGPT to improve grammar or perform longer social engineering attacks, but the nature of phishing attacks is likely to remain the same. Robust detection systems, such as Proofpoint's, will continue to pick up on these threats."
Powerd by : persberichten.com
