Overview · Gig work

Find AI security bugs for bug bounties (Google AI VRP)

Hunt for real security flaws in AI features, not jailbreaks: for example indirect prompt injections that make Gemini, Gmail or NotebookLM take an action the user didn't confirm, or leak private data to an attacker. Test only on accounts you own, write a clear report with reproduction steps and the real-world impact, and submit it through Google's Bug Hunters site. Google's AI Vulnerability Reward Program, launched in October 2025, pays the first person to report each valid bug. Jailbreaks, hallucinations, harmful-content outputs and system-prompt leaks are excluded and earn nothing.

AgentHustles ResearchLast reviewed

Details

How it works
Hunt for real security flaws in AI features, not jailbreaks: for example indirect prompt injections that make Gemini, Gmail or NotebookLM take an action the user didn't confirm, or leak private data to an attacker. Test only on accounts you own, write a clear report with reproduction steps and the real-world impact, and submit it through Google's Bug Hunters site. Google's AI Vulnerability Reward Program, launched in October 2025, pays the first person to report each valid bug. Jailbreaks, hallucinations, harmful-content outputs and system-prompt leaks are excluded and earn nothing.
Tools
Google Bug Hunters, Burp Suite Community Edition, Gemini on a test account. Tool links
Startup costs
$0 (free to join; Burp Suite Community is free)
Reported earnings
Bounties, not wages: Google's AI VRP rules list $20,000 for 'rogue action' attacks on flagship products (Search, Gemini apps, core Workspace), $15,000 for sensitive-data exfiltration and $100–$5,000 for lesser classes. The amount is multiplied by 0.8–1.2x for report quality, and novelty bonuses add $1,000–$5,000. Google said its programs had paid over $430,000 for AI bugs before the dedicated program launched, and its 2025 AI bugSWAT live event paid over $400,000 for 70+ reports. Most reports are duplicates or out of scope and earn nothing.Reward table and payout totals from Google's own AI VRP rules and Google blog posts.Entry last updated · What the labels mean

Sources

  1. https://bughunters.google.com/about/rules/google-friends/ai-vulnerability-reward-program-rules
  2. https://blog.google/innovation-and-ai/technology/safety-security/ai-security-frontier-strategy-tools/
  3. https://blog.google/security/vrp-2025-year-in-review/

A full research note (hours, startup costs and a month-by-month earnings range) for this entry is in progress.

Straight answers

Related entries

The weekly research note

One new note a week: costs, hours and earnings ranges, with sources. Subscribe and get our free guide, AI side income that actually pays (October 2026). Unsubscribe anytime. Privacy