A new report released today by cloud cybersecurity firm Barracuda Networks Inc. finds an alarming rise in email-based attacks, with cybercriminals increasingly leveraging generative artificial intelligence to craft more convincing and sophisticated phishing and business email compromise attacks.

The Top Email Threats and Trends Report, based on the analysis of 69 million email-based attacks across 4.5 million mailboxes over a period of 12 months, found that scamming and phishing emails now make up 86% of social engineering attacks. BEC attacks account for 10.6% of these attacks, with conversation hijacking incidents also increasing by 70% since 2022.

The report runs through various methods and types of attacks. QR code phishing came to the fore in 2023, with one in 20 mailboxes targeted in the last quarter of 2023. Through QR Codes, the attacks often bypass traditional email filtering methods, making them particularly risky.

The most popular email service targeted by cybercriminals was Google LLC’s Gmail, with 22% of all observed social engineering attacks targeting users of the service. On the one hand, Gmail is highly popular and cybercriminals are looking to maximize their potential reach, but Gmail also has some of the best email protection tools in the market.

The report also delves into the use of URL-shortening services such as bit.ly, which were found to be used in nearly 40% of social engineering attacks. Using a URL shortener obscures the true nature and destination of malicious links, making it easier for attacks to deceive their victims.

On generative artificial intelligence, the report highlights how, since the launch of ChatGPT in late 2022, cybercriminals have been leveraging these technologies to automate the creation of phishing, spear phishing and business email compromise attacks.

The use of generative AI by attackers allows them to craft highly personalized and contextually relevant messages, increasing their chances of success. As AI-generated emails are typically free of grammatical errors, the added layer of sophistication makes them harder for traditional security measures to detect.

Additionally, cybercriminals are using fine-tuned AI systems accessible via the dark web, such as WormGPT and DarkBERT, to generate malicious code, gather open-source intelligence and tailor their attacks to specific targets.

The report recommends multilayered email security measures, such as deploying advanced AI-powered security technologies, implementing multifactor authentication, automating incident response and enhancing cybersecurity awareness through regular training and phishing simulations, to provide the best protection against email attacks of all forms.

“As cybercriminals continue to adapt their tactics, IT and security professionals need to stay focused on the evolution of email attacks and the influence generative AI has on these types of threats,” the report notes.

Image: SiliconANGLE/DALL-E 3