AI security risks are changing how MSPs protect the inbox
AI gives attackers flawless grammar, branding and tone, while your clients' own AI tools can quietly expose confidential data. Both sides of that problem land in the inbox you protect.
A two-sided problem
Channel Insider covered the AI security conversation at GTIA ChannelCon 2026, in reporting by managing editor Victoria Durgin. Attackers use AI to make old attacks more convincing, and customers adopting AI tools are exposing confidential data. Mailprotector Founder and CEO David Setzer called AI "human intelligence at scale."
"AI creates the perfect phishing message, with perfect grammar, perfect colloquialisms and perfect fake logos. If you look at phishing messages now, you cannot distinguish them from a message from a real company."
David Setzer, Founder and CEO, Mailprotector
The report also features MJ Shoer of GTIA, MSP founder and Wake Technical Community College professor Sandellyo Kauba, and Demetrios Georgiou of Coro.
What MSPs are asking
- Can users still spot AI phishing? Not reliably. The typos and odd phrasing awareness training relied on no longer show up.
- What stops it, then? Changing the default. Zero trust email security treats no message as automatically trustworthy and makes every one earn its way to the mailbox.
- Does "AI-powered" security scan every message with AI? Not at scale. The compute cost makes it impractical, and many of those claims describe machine learning that has been in filters for years.
Intentional AI in Shield
Shield's zero trust model controls who reaches the inbox. AI supports specific features on top of that foundation, such as summarizing messages in Bundler so users can quickly decide what deserves a closer read.
Check a client's posture
Radar runs live round-trip email tests and returns a risk score at no cost.