Vulnzero is marketing. But that doesn’t mean your board should ignore this warning. Will they hear it from Anthropic or your security leader? Or from the news coverage of the coming wave of breaches and hacks?
Models are getting better and faster at finding bugs. Most risk sits in legacy systems, vendor products, OT and embedded devices, unmaintained code, misconfigurations, poor practices, and process failures.
Discovery is not the constraint and never has been. Remediation capacity and prioritization is. We continue pushing vulnerabilities into prod at high rates. Configuration still misses quite often. The volume of findings will rise faster than most organizations can prioritize and fix them.
Anthropic framing the next 18 months as a temporary crisis before permanent resolution is useful for product positioning. It is not an operational forecast.
It is going to get uglier. Attack surface reduction works. Do the basics. Reduce exposure. Improve remediation. Stop treating how you communicate and use data as a playground. Treat AI discovery as an accelerator of existing work, not the end of it.