Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

AI agents have crossed an important line from making suggestions to taking actions. They can read a repository, call internal systems, change code, open a pull request, and keep working while the human moves on.

It happened again. We blinked, and suddenly summer’s over and it’s time to register for classes. The horror!

AI is changing how products get made. For user experience teams, that means the very shape of the work is changing.

It's 6:30am and you hear the door of the nightclub you've spent the last 8 hours inside shriek as it closes behind you. You watch bleary-eyed as an overly bright sunrise illuminates the business-suited people as they glide effortlessly along the sidewalk, their obnoxiously well-rested faces talking about work on their fully charged phones. You wonder, "Where did all the fun people go? And what happened to my wallet?".

We studied what happens when Large Language Models (LLMs) generate vulnerability patches for recently disclosed, complex vulnerabilities. Our data shows that LLMs produce Fix-Like Artifacts with Embedded Defects (FLAWED) 53.9% of the time when complex patches are required.

AI has changed the calculus of a credential attack. Before, finding and exploiting credentials in an enterprise environment required time, patience, and human judgment. An attacker had to decide which accounts were worth testing and which systems were worth reaching. Many credentials never made the list.

On August 2, 2026, Productiv told customers its SaaS management platform was shutting down on August 6, with account data deleted once access ended. Four days is not much time to pull years of app inventory, spend, and usage data out of a system you've come to depend on, especially with AI tools now adding a fast-moving new layer of spend and access to track on top of everything else.

Earlier this year, a bill arrived from one of 1Password’s AI vendors for 5x the value of the original contract. The initial agreement came in below a certain threshold, so it never reached the right approvers for review. By the time it did, we had a much clearer understanding of how quickly AI costs can add up.

AI agents are doing more than just generating code. Increasingly, they are working autonomously on complex coding challenges, touching production APIs, databases, and infrastructure across development environments, often without thorough human review. To perform these operations and access multiple systems, agents rely on developer secrets and non-human identities (NHI). But often, developers lack a secure way to share these secrets, leading to overprivileged, invisible access.

In our last post, we shared how we began to scale our security code review process with SAGE. We discussed how we gathered historical Product Security (ProdSec) review records to create a 1Password-specific ruleset, the three-stage Finder/Critic/Judge pipeline, and the limitations of our v1 implementation.