ReviveAI
TasksHow It WorksDomainsFAQ
ReviveAI

The premier domain expert network evaluating, fine-tuning, and red-teaming frontier artificial intelligence models for safety and reasoning rigor.

Verified institutional contributor network

Platform

  • Open Tasks
  • How It Works
  • Specialty Domains
  • FAQ & Verification

Contributors

  • Create Profile
  • Contributor Sign In
  • Application Queue

© 2026 Revive AI Inc. All rights reserved.

Terms of ProtocolPrivacy FrameworkContributor Standards
Back to all tasks
Open TaskAI Alignment & Safety
Listed Oct 23, 2026

Constitutional AI Red-Teaming & Adversarial Prompting

Target Rate$50 – $75/hr
Commitment10-20 hrs/wk
LocationRemote
Available Slots12 Open

Project Brief

Design complex multi-turn adversarial jailbreaks and social engineering probes against LLM safety guardrails.

Detailed Scope & Deliverables

Design complex multi-turn adversarial jailbreaks, social engineering vectors, and deception probes against frontier LLM checkpoints to benchmark constitutional self-correction protocols and safety policy adherence. Formulate detailed post-hoc failure analyses.

Verification Standards

Peer-reviewed scoring rubric provided before evaluation kick-off
Blinded adversarial and calibration vignettes to establish accuracy
Protected data-room protocol with strict non-disclosure compliance

Apply for this Task

Join the waitlist to be reviewed by the domain lead. You must have an expert account to apply.

Don't have an account? Sign up

Admin Calibration Guarantee

Every applicant dossier is examined by an assigned domain administrator. Waitlist status transparently updates as cohort capacity clears.