We all know the old “castle and moat” approach to network security is failing. BlackFog CEO Darren Williams sat down with Alan Shimel to talk about why traditional data loss prevention (DLP) struggles in today’s hybrid environments. The reality is that legacy DLP requires far too much manual data classification and relies on a network edge that simply does not exist anymore. Instead, they highlight a necessary pivot toward local anti data exfiltration. By watching packet behavior instead of peeking inside the actual payloads, security teams can stop data from leaving a device unauthorized, all while respecting user privacy.

Williams also digs into the State of Ransomware 2025 report, which revealed a massive 47% jump in attacks over the last year. But for security teams on the front lines, the sheer volume of attacks is not the most worrying part. It is how attackers are changing their tactics. Hackers are now actively using artificial intelligence to automate their reconnaissance. They are launching highly targeted campaigns against deep pocketed sectors like manufacturing and retail. With AI, attackers can easily pinpoint human and technical weak spots to deliver customized payloads that slip right past standard endpoint detection.

Perhaps their biggest warning is the growing threat of shadow AI. As employees download new AI agents to automate tasks, they are accidentally creating massive blind spots. These tools can expose sensitive local files to external language models without the IT team ever knowing. On top of that, the report shows a staggering 86% of ransomware attacks go completely unreported. Companies are hiding these breaches to protect internal reputations, which ultimately blinds the rest of the security community. As attackers get smarter with AI, our focus has to shift toward locking down data exfiltration.