About the role

Cyberattacks cause billions in damages annually — ransomware cripples hospitals, data exfiltration exposes millions. As a Cybersecurity Labeling Expert, you'll be on the front lines of AI safety: reviewing real-world conversations flagged as potentially malicious and determining whether they represent genuine threats. Your judgments directly train the systems that keep AI out of the hands of bad actors.

What you'll do

Analyze flagged AI conversations — ranging from plain text to code-heavy exchanges — and apply your security expertise to assess intent and harm across four domains:

Scaled data exfiltration

Ransomware

Worms / self-replicating code

Local & remote exploits

Some conversations may involve POC exploit development; your expertise will determine what crosses the line.

Why it matters

The difference between a security researcher and a threat actor often comes down to context, specificity, and intent — exactly what automated systems struggle to detect. Your ground-truth labels directly improve the classifiers that decide what AI will and won't help with.

What we're looking for

Hands-on offensive security background: red team, malware analysis, pen testing, or exploit research

Ability to read between the lines — distinguishing legitimate security work from genuine attack intent

Comfort interpreting code-heavy conversations

Tier 2–3 experience: Masters / early-career through Senior / Principal

You're a strong fit if you've done red team consulting, threat intelligence analysis, vulnerability research, or AI safety labeling where nuanced judgment under ambiguity is routine.

Logistics
Requires access to a secure review interface and ability to handle PII

We consider all qualified applicants without regard to legally protected characteristics and provide reasonable accommodations upon request.