Security Weekly

Everyone talks about prompt injection.

But this security discussion argues that's not the biggest AI risk.

The real question is what an AI agent is allowed to do when it has access to systems, tools, and permissions. Old security problems may be showing up in new AI-powered forms.

Are companies focusing on the wrong AI threat?

Anomaly detection never really disappeared.
Now it’s being rebuilt using AI architectures similar to large language models.

Instead of looking at isolated alerts, these systems analyze long behavioral sequences across logs, DNS, WAF traffic, and network flows to identify suspicious patterns.

Could AI finally make anomaly detection useful at scale?

AI can process enormous amounts of security data faster than humans ever could.

But when it comes to context, intuition, and seeing the bigger picture, humans still fill critical gaps that automation struggles with.

Is the future of cybersecurity really humans versus AI — or humans working with AI?