Being recognized as AI Security Innovator of the Year

I was recognized as AI Security Innovator of the Year in The Hacker News Cybersecurity Stars Awards.

That is a nice sentence to write. I am not going to pretend otherwise.

The award points at work I have been building at Keeper around AI-native threat detection, agent governance, privileged sessions, and keeping credentials out of developer workflows. The title matters less to me than the engineering problems underneath it.

AI security gets real when it changes the system:

• what an agent can see
• what it can change
• whose authority it uses
• where secrets are resolved
• when a person approves a step
• what evidence survives afterward

Those questions show up in live session monitoring, MCP integrations, agent tooling, and developer workflows. They are different surfaces with the same pressure: a conversational interface can make broad authority feel casual.

The safer pattern is fairly plain. Give agents scoped grants. Keep credentials in hardened storage instead of chat. Record the initiator, executor, authority, and side effect. Watch privileged behavior while the session is alive rather than treating detection as an after-action report. Give analysts a way to understand why a signal appeared and what the system missed.

The controls have to fit the workflow. If the safe path is too slow, people route around it. If approval copy is vague, people click through it. If logs are unreadable, incident review becomes archaeology. If secret handling is painful, somebody pastes the key into the prompt.

That is the work I want to keep building: safer defaults that still let people get their jobs done.

Awards are snapshots. The systems keep changing. Agents are becoming part of the security surface, and the useful answer is not hype or fear. It is architecture that makes access narrow, credentials private, approvals meaningful, and evidence readable.