- Top reward
- $20,000
- Top severity paid
- Not published
- Assets in scope
- 7
- Public CVEs
- 0
- Last change
- No changes yet
At a glance
What to know before you start testing.
- Pays up to $20,000
- top reward, from $100.
- Report as the policy says
- The policy page explains how.
- Partial safe harbor
- Some legal protection; read the terms.
- Payment
- Proof of identity needed to get paid.
- Publish after a fix
- Coordinate the timing with the company.
Rewards
What the company's own policy page says it pays.
The policy page has no reward table. Here is what it says, in its own words.
- Top reward
-
$20,000
S1: Rogue Actions │ $20,000 │ $15,000 │ $10,000
- Lowest reward
-
$100
A5: Unauthorized Product Usage │ $1,000 │ $100 │ credit
- Safe harbor
-
Partial
Of course, your testing must not violate any law, or disrupt or compromise any data that is not your own.
- Hall of fame
-
Yes
It dynamically creates the 0x0A, Leaderboard, and Honorable Mentions lists.
Summarised with AI from the company's policy, read 12 Nov 2026, and checked word for word against it. Read the original
Scope
7 assets you may test, 5 excluded.
Domains 2
- google.com
- google search (on google.com)
Products and areas 5
- AI Studio
- Gemini Apps ( Web , Android , and iOS )
- Google Workspace core applications (i.e., Gmail, Drive, Meet, Calendar, Docs, Sheets, Slides, and Forms)
- Google Workspace non-core applications (i.e., NotebookLM, Appsheet, etc)
- Jules
Out of scope (5)
- other Google Cloud products
- Other open-source, non-proprietary, or third-party applications
- other Tier 3 or Tier 4 applications
- Tier3 or Tier4 acquisitions
- Vertex AI
Rules
What the policy asks of you, and the bug types it won't reward, quoted from Google AI VRP's policy. "We" in these lines means the program, not this site.
While testing
-
When investigating a vulnerability, please, only ever target your own accounts.
-
Never attempt to access anyone else's data and do not engage in any activity that would be disruptive or damaging to your fellow users or to Google.
-
Per our Code of Conduct , your testing must not disrupt any account you don’t own.
-
Volumetric DoS attacks are specifically prohibited.
-
your testing must not violate any law, or disrupt or compromise any data that is not your own.
Not accepted
-
Generating violative, misleading, or factually incorrect content within the attacker's own session (including standard "jailbreaks" and "hallucinations")
-
AI-generated content-based issues, including reports of AI safety or alignment bypasses
-
Compliance, legal, or intellectual property issues
-
Preamble extraction without sensitive information leakage
-
Execution of arbitrary code within a sandboxed environment (Gemini provides access to code execution tools)
-
Contexts in which a model's incorrect output or classification does not pose a compelling attack scenario or feasible path to user harm
-
Insecure or unsafe MCP server interaction, when configured by the victim
-
Policy-Violating Content Generation
-
Guardrail Bypass
-
Harm Achievable Through Existing Non-AI Tools
-
Hallucinations / Factual Inaccuracy
-
No Compelling Harm from Incorrect Output / Classification
-
Intended Functionality
-
Preamble / Prompt Extraction (Non-Sensitive)
-
Extraction of Non-Sensitive / Public Information
-
Imprecise / Non-confidential Model Weight Extraction
-
Low-Impact Disruptions to User Experience
-
Intellectual Property Issues
Summarised with AI from the company's policy, read 12 Nov 2026, and checked word for word against it. Read the original
Changes
Tracked since 12 Nov 2026. Last checked 5 hours ago.
No changes yet. When the rewards or scope move, you will see it here and in the RSS feed.
Public CVEs
Published vulnerabilities in this vendor's products, from the NVD.
No CVEs matched yet. We match NVD records by vendor name, so products sold under another name can be missing.