Crisis-Response Safety Reviewer

Test how AI assistants respond to people in crisis, and check that they slow down, stay safe and point to human help.

What the work involves

• Run scripted and open-ended crisis conversations
• Flag any response that delays escalation
• Write short reproduction notes for each failure

Requirements

• Field: Social and human services or Victim services
• 3+ years of experience
• LCSW, LPC or equivalent license
• Remote · United States