Free tool
AI Tool Risk Directory
Every AI tool we rate, in one place, alphabetical, with its plain-English risk rating and the tiers that change it.
Independent AI-leakage risk ratings for the AI tools small businesses actually use. Each tool is rated 1 (lowest risk) to 5 (highest risk), the chance that sensitive data leaves your control through the tool. Lower is safer. Ratings reflect our editorial judgement from documented evidence, not vendor input.
Compare by type: chat assistants, meeting note-takers, coding tools, writing, design and voice, team work tools, customer service and CRM, accounting and online selling.
Read the rating column carefully: most tools are rated differently across their plan tiers. A consumer or free tier is often higher-risk than the same vendor’s business or enterprise tier, which may carry no-training or zero-retention guarantees. Where a tool has a split rating, both are shown. Click any tool for the full breakdown, the plan-by-plan detail, and sources.
| AI tool | Category | Risk rating (lower = safer) |
|---|---|---|
| Adobe Firefly | Vertical SaaS AI | 1 of 5 (Firefly Enterprise with indemnification) / 2 of 5 (Creative Cloud Firefly and Firefly Free) |
| Asana AI / Asana Intelligence | Embedded Productivity AI | 2 of 5 |
| Canva Magic Studio | Vertical SaaS AI | 3 of 5 (Free/Pro) / 1 of 5 (Teams/Business/Enterprise/Education with Canva Shield) |
| ChatGPT (OpenAI) | Direct AI Tools | 3 of 5 (Free/Go/Plus/Pro) / 2 of 5 (Business, Enterprise, and API with zero data retention) |
| Claude (Anthropic) | Direct AI Tools | 3 of 5 (Free/Pro/Max) / 2 of 5 (Team, Enterprise, and API) |
| Cursor (Anysphere) | Direct AI Tools | 3 of 5 (Hobby/Pro without Privacy Mode) / 2 of 5 (Pro with Privacy Mode) / 1 of 5 (Teams/Enterprise) |
| ElevenLabs | Direct AI Tools | 4 of 5 (Free/Starter/Creator) / 3 of 5 (Pro/Scale) / 2 of 5 (Enterprise with custom DPA) |
| Fathom (meeting AI) | Embedded Productivity AI | 3 of 5 (Free/Premium) / 2 of 5 (Team Edition with custom contract) |
| Gemini (Google) | Direct AI Tools | 3 of 5 (consumer Gemini) / 1 of 5 (Workspace Gemini) |
| GitHub Copilot (GitHub / Microsoft) | Direct AI Tools | 4 of 5 (Free/Pro/Pro+/Max) / 2 of 5 (Business/Enterprise) |
| Grammarly (with Grammarly AI / GrammarlyGO) | Vertical SaaS AI | 3 of 5 (Free/Premium) / 2 of 5 (Business/Enterprise) |
| Granola | Embedded Productivity AI | 4 of 5 (Free/Business, training on by default and no disclosure to other participants) / 2 of 5 (Enterprise Workspace, training off by default) |
| Grok (xAI) | Direct AI Tools | 5 of 5 (Consumer Grok) / 3 of 5 (Grok API) |
| HubSpot AI (Breeze) | Vertical SaaS AI | 2 of 5 |
| Intercom Fin | Vertical SaaS AI | 2 of 5 |
| Jasper (Jasper AI) | Direct AI Tools | 3 of 5 |
| Linear AI | Embedded Productivity AI | 2 of 5 |
| Microsoft Copilot | Direct AI Tools | 2 of 5 (Microsoft Copilot on a personal Microsoft account, current app and web) / 2 of 5 (Microsoft Copilot for work or school, formerly Microsoft 365 Copilot, Entra sign in). Legacy consumer app not yet updated: 3 of 5, see the training section. |
| Midjourney | Direct AI Tools | 4 of 5 (Basic/Standard) / 3 of 5 (Pro/Mega with Stealth Mode) |
| Notion AI | Embedded Productivity AI | 2 of 5 (Free/Plus/Business) / 1 of 5 (Enterprise with zero-retention APIs) |
| Otter.ai | Embedded Productivity AI | 5 of 5 (Free/Pro pending litigation outcome) / 3 of 5 (Business/Enterprise) |
| Perplexity | Direct AI Tools | 3 of 5 (Consumer Perplexity) / 2 of 5 (Enterprise Perplexity) |
| Replit (with Replit AI) | Direct AI Tools | 5 of 5 for production environments / 3 of 5 for prototyping only |
| Salesforce Einstein / Agentforce | Vertical SaaS AI | 2 of 5 |
| Shopify Magic / Sidekick | Vertical SaaS AI | 3 of 5 |
| Slack AI (Salesforce) | Embedded Productivity AI | 3 of 5 |
| Xero (with JAX / Just Ask Xero) | Vertical SaaS AI | 2 of 5 |
| Zendesk AI (Advanced AI / Resolution Platform) | Vertical SaaS AI | 2 of 5 |
| ZoomMate | Embedded Productivity AI | 2 of 5 |
Ratings are reviewed as vendors change their data policies and as new incidents come to light. This directory is the free summary layer; the full profiles carry the retention terms, training-opt-out steps, breach history, and the evidence behind every rating.
How this was written: this page was researched and drafted with AI assistance (primarily Claude, made by Anthropic) and checked against primary sources before publication. We hold ourselves to the same standard we rate other tools against, see How This Site Uses AI for the full disclosure.