Proofpoint principal threat researcher Selena Larson highlights a phenomenon where AI is making cyber activities less sophisticated, and the most effective threat actors are avoiding it entirely This article explores ai making threat. . Related: F-Secure Launches Victim-First Scam Encyclopedia Covering 50+ Types of Consumer Fraud There's a lot of pressure to showcase impressive things, but I strongly disagree with the narrative that AI is making threat actors better.

According to Larson, groups like TA2725, a former player in the Brazilian banking Trojan scene that has now degenerated into a spam factory, are using AI to increase the volume of their attacks while decreasing their sophistication. Since March, TA4903 has utilized AI across all stages of its attack chain: social engineering lures, targeting, and credential harvesting.

They are now sending victims QR codes that don't display in emails, creating sophisticated PDFs that are attached to empty emails, and crafting emails that impersonate legitimate organizations but do not replace default, auto-generated AI content. For instance, because it was so easy to create, APT36 could overwhelm its targets with an abundance of malware, making it nearly impossible to significantly impact just one or two implants. However, in terms of effectiveness, deploying an AI-generated Go-based malware tool that couldn't exfiltrate data or a bot management system that didn't function as intended was not particularly useful, according to cyber researchers.