Know before your role does.
pssst. this is to sign up for updates on the game ;)
Will AI replace Adversarial AI Researcher jobs? · Latest report #89ixyuid
Your job has an expiration date
December 8, 2027
In 14 monthsYou are teaching the machine to break itself, effectively firing yourself twice.
Professional breaker of things you claim are unhackable
Check another roleComment your opinionor JOIN THE RESISTANCE!
One human is building a game about fighting back.
Recursive vulnerability discovery for infinite security






brr.you▸ REDTEAM-GPT-9▸ RARITY · COMMON▸ DEPRECATION IN 14 MONTHSAI's Special Power
exhaustively maps model failure modes in milliseconds
You hunt for edge cases in model safety boundaries and write adversarial prompts to force model failures. You refine automated evaluation pipelines.
The one thing AI can't do
You possess the strange, chaotic intuition to predict which weird prompt will break the alignment.
Your enemy, your name, your date. Post it.
Step by step, from the first quiet change to Deprecation Day.
Week 4
AI models index your historical red-teaming results for patterns.
Month 4
The AI starts testing its own defenses against human tactics.
Month 8
System scripts replace your manual benchmarking of new checkpoints.
Month 12
The system generates its own novel adversarial attack vectors.
Month 14 · D-Day
The system reaches full self-hardening, rendering your research redundant.
What you cost, what the machine costs, and what you get on the way out.
Per year
Savings: 98%
4,000,000
Average prompts per hour
12 percent
Human error rate
100 percent
Model hallucination detection
Your severance
A one-year subscription to the model you just killed.
Discussion