PARTNERS

Add selected Workbench capabilities through bounded OEM and partner integrations

Flagship research · 2026

Published

What companies are actually hiring for when they say AI Security Engineer.

A ground-truth study of public hiring signals, capability expectations, role compression, and the operating-model evidence behind an emerging security discipline.

754,507

Job descriptions analyzed

27,260

Companies analyzed

15

Chapters

9

Role archetypes mapped

Free PDF · No signup required

The market signal

One emerging role. Several jobs hiding inside it.

The report turns a large corpus of job-description intelligence into a practical view of the capabilities employers name, combine, omit, and struggle to separate.

Finding 01

The Frankenstein Role

Public hiring signals increasingly compress application security, cloud security, ML security, governance, and product engineering into one role.

Finding 02

Skill Washing

Putting AI in a title is not the same as specifying AI security work. The report separates explicit capability signals from adjacent language.

Finding 03

The Unicorn Index

Some job descriptions appear to price one role while describing team-shaped capability across several security and engineering disciplines.

Finding 04

The Evidence Gap

Governance language becomes operational only when it reaches controls, telemetry, approvals, remediation, and evidence.

Inside the report

Research made legible as systems, boundaries, and operating decisions.

Role compression figure from The State of AI Security Engineering Report

Role compression

How distinct security and engineering responsibilities converge in AI security hiring language.

The authority ladder figure from The State of AI Security Engineering Report

The authority ladder

A practical model for reasoning about systems that move from generation toward delegated action.

The MADE operating model figure from The State of AI Security Engineering Report

The MADE operating model

A shared vocabulary for mapping, attacking, defending, and evidencing AI systems.

Research standard

Built to inform decisions without overstating what public data can prove.

Corpus

Public hiring signals

The analysis treats job descriptions as evidence of market demand and role language—not as proof of company security maturity.

Interpretation

Aggregate benchmark

Patterns are evaluated across the corpus so the report can describe directional labor-market signals without accusing individual companies.

Claim discipline

Evidence before assertion

Findings are separated from interpretation, and claims are bounded by explicit caveats and source notes.

Based on analyzed job-description signals, not proof of any individual company's internal security maturity.