The cybersecurity AI war just got real. While everyone's been talking about agents replacing workers, the actual battle is happening in security teams: AI attackers versus AI defenders, and the defenders just got some serious new weapons.
01
Claude Security goes live: point, scan, fix
Anthropic launched Claude Security in public beta, built directly into Claude Code. Security teams can point it at any repository, get validated vulnerability findings, and fix them in the same interface where they write code. No custom tooling, no API integration required.
Why it matters: This is security scanning that actually fits into developer workflow instead of disrupting it. Every startup that's been putting off proper security audits because the tooling is too complex just lost their excuse.
Box CEO Aaron Levie predicts the end of software UIs as we know them
Box CEO Aaron Levie outlined how the software business model changes when AI agents become the primary users of most applications. His thesis: agents won't use user interfaces, they'll talk directly to APIs. This means every software company needs a "headless" version of their product, but the pricing models around API-first usage are still being figured out.
Why it matters: If you're paying per-seat for software today, that pricing is about to change completely. When one human can manage ten AI agents, and those agents are the ones actually using the software, your SaaS bills are going to look very different.
Cybersecurity leaders say AI is both the problem and the solution
Databricks founder Aditya Agarwal interviewed Palo Alto Networks CEO Nikesh Arora about the new reality of cybersecurity: attackers have AI, so defenders need AI too. The conversation covered how AI is fundamentally reshaping both sides of the security equation, with AI becoming both the primary threat vector and the only viable defense.
Why it matters: Your company's security team is about to start an arms race they can't opt out of. The attacks coming in 2026 will be generated by AI, which means human-speed defenses won't work anymore.
PSPDFKit founder Peter Steinberger ships security collaboration with tech giants
PSPDFKit founder Peter Steinberger shared details about an extensive security ecosystem collaboration involving Nvidia, OpenAI, Microsoft, GitHub, Tencent, Convex, Atlassian, and Blacksmith. The project appears focused on securing AI development tools and infrastructure.