Aaron Levie: Token Traffic from Cybersecurity Agents May Grow 1,000x
Box CEO Aaron Levie predicts that before long, the number of tokens flowing through cybersecurity agents will be 1,000 times greater than it is now. He believes that as the attack surface expands, enterprises will struggle to keep up with the volume of new risks, while swarms of agents can invoke various security tools, test attack variants, read logs, and assess enterprise data risks. This capability stems from the maturation of long-running agents and computer use over the past few months, as well as deep model training in the cybersecurity domain.
View original postNode2Shell (CVE-2026-111497) is a memory bug in Brotli, the compression library bundled into Node.js, Python, system libs, Envoy and more. Most teams don't know every place Brotli runs in their stack. It's usually a transitive dependency or an OS package, and pip-audit may not flag it yet. Enclave's agents scan your code, cloud and servers and show you exactly where you're exposed. Advisory + free check: https://t.co/n2WDOnYAj9
View original
Node2Shell (CVE-2026-111497) is a memory bug in Brotli, the compression library bundled into Node.js, Python, system libs, Envoy and more. Most teams don't know every place Brotli runs in their stack. It's usually a transitive dependency or an OS package, and pip-audit may not flag it yet. Enclave's agents scan your code, cloud and servers and show you exactly where you're exposed. Advisory + free check: https://t.co/n2WDOnYAj9
Body
We’ll likely have 1,000X more tokens flowing through cyber agents soon than we do today.
As the surface area of cyber risk increases, at some point soon it will be nearly impossible to keep up with the volume of new risks (we’re essentially already there). Large companies barely had the resources to pull this off before, but the vast majority of companies don’t.
There’s a never ending set of libraries to keep up with, tools to leverage, and sprawling infrastructure and code to secure. We’ll have agent swarms that can fan out, leverage any security tool, test various perpetuations of attacks, read through logs, process enterprise data for risk level, and more.
This capability only emerged in the past few months. Wouldn’t have been possible before long running agents and computer use were good enough, and models hadn’t been trained on cyber deeply. That’s all changing now.
This is a huge area of opportunity right now for every layer of the security stack across identity, firewall, network, code scanning, data protection, and more. And it’s at least one reason to be optimistic about the balance of offensive and defensive capability.