Internet egress is outbound network access from a controlled system to the wider internet. In practice, it is the path that lets a server, sandbox, container, or AI agent make external DNS queries, open TCP or UDP connections, download resources, call APIs, or send telemetry beyond its local boundary.
Security teams care about egress because it is often the point where a contained workload can reach real services. Attackers abuse it for command-and-control traffic, data exfiltration, proxying, and beaconing. Defenders use egress controls such as allowlists, firewalls, NAT gateways, DNS filtering, logging, and network segmentation to limit where systems can go and to detect unexpected outbound traffic. In testing environments, uncontrolled egress can turn a simulation into live network contact, making results less reliable and increasing the risk of touching production infrastructure by mistake.