Most (87%) safety professionals have reported that their group has encountered an AI-driven cyber-attack within the final 12 months, with the expertise more and more takes maintain, in keeping with a brand new report by SoSafe.
The brand new SoSafe 2025 Cybercrime Traits report additionally famous that 91% of all safety specialists anticipate a major surge in AI-driven threats over the subsequent three years.
The World Financial Discussion board’s International Cybersecurity Outlook 2025 cited a 223% enhance within the commerce of deepfake-related instruments on darkish net boards between Q1 2023 and Q1 2024.
The SoSafe examine of 500 world safety professionals, in addition to 100 SoSafe prospects throughout 10 international locations, famous that detection of those assaults continues to be a problem, with solely 26% expressing excessive confidence of their detection skills.
Regardless of this, 96% of respondents acknowledged the significance of detecting AI-based assaults.
Andrew Rose, CSO, SoSafe, commented, “AI is dramatically scaling the sophistication and personalization of cyber-attacks. Whereas organizations appear to pay attention to the menace, our knowledge reveals companies should not assured of their means to detect and react to those assaults.”
How AI is Leveraged in Cyber-Assaults
Obfuscation methods, akin to AI-generated strategies to masks the origins and intent of assaults, have been cited as the highest concern by 51% of safety leaders.
The analysis additionally famous that AI can also be enabling multichannel assaults which mix ways throughout electronic mail, SMS, social media and collaboration platforms.
Over the previous two years, 85% of 95% of cybersecurity professionals agree they’ve observed a rise on this model of assault.
“Focusing on victims throughout a mix of communications platforms permits them to imitate regular communication patterns, showing extra official,” stated Rose. “Simplistic electronic mail assaults are evolving into 3D phishing, seamlessly integrating voice, movies or text-based components to create AI-powered, superior scams.”
SoSafe additionally famous that the in-house adoption of AI is inadvertently increasing organizations’ assault surfaces, subjecting themselves to new progressive assaults akin to knowledge poisoning and AI hallucinations.
The survey discovered that 55% of companies haven’t totally applied controls to handle the dangers related to their in-house AI options.
SoSafe famous that AI does have the potential to be “transformational” in cybersecurity protection by way of worker coaching, correlating safety alerts and automating code corrections.
“Whereas AI undoubtedly presents new challenges, it additionally stays one in all our biggest allies in defending organizations in opposition to ever-evolving threats. Nonetheless, AI-driven safety is barely as robust because the individuals who use it. Cybersecurity consciousness is vital. With out knowledgeable staff who can acknowledge and reply to AI-driven threats, even the perfect expertise falls quick. By combining human experience, safety consciousness and the cautious utility of AI, we will keep forward of the curve and construct stronger, extra resilient organizations,” stated Niklas Hellemann, CEO of SoSafe.
