AI Privacy Scorecard 2025: 8 AI Apps and Their Hidden Privacy Risks

Which AI apps truly protect your privacy? Our comprehensive scorecard reveals all

Your conversations with AI apps aren't as private as you think. A new report from Incogni evaluates the data privacy practices of today's most widely used AI platforms, revealing shocking differences in how these companies handle your personal information.

The goal is to demystify these often complex documents, helping you understand how your data is handled when you interact with these AI assistants, because what you don't know could cost you. Here's our comprehensive privacy scorecard, ranked from worst to best:

The Privacy Rankings

1. Lensa AI

Lensa AI

3/10

POOR PRIVACY

Lensa's viral success with AI avatars came with privacy trade-offs. The app processes uploaded photos to generate avatars, and past controversies revealed unclear retention policies.

Source: Lensa AI Privacy Policy | TechCrunch

2. Meta AI

Meta AI

4/10

POOR PRIVACY

Meta's approach reflects broad cross-platform integration (Facebook, Instagram, WhatsApp). This enables personalization but raises privacy concerns for sensitive conversations.

Source: Meta Privacy (blog) | Meta Privacy Policy

3. Google Gemini

Google Gemini

5/10

FAIR PRIVACY

By default, prompts and conversations may be used to improve models. Activity is tied to your Google Account and retention settings can vary by plan.

Source: Google Gemini Help | Google Privacy Policy

4. OpenAI ChatGPT

OpenAI ChatGPT

6/10

AVERAGE PRIVACY

ChatGPT is relatively transparent, but consumer data may be used to improve models unless you change settings or use paid/enterprise options with explicit guarantees.

Source: OpenAI Privacy Policy | OpenAI Data Controls

5. Hugging Face

Hugging Face

6/10

AVERAGE PRIVACY

Open ecosystem with strong self-hosting options — privacy depends on the endpoints you choose and contract terms. Self-hosting = best privacy.

Source: Hugging Face Privacy | Hugging Face Security

6. Canva AI

Canva AI

7/10

GOOD PRIVACY

Canva states user content isn't used to train AI unless you opt in. Privacy toggles are available in account settings for creators.

Source: Canva Privacy Policy | Canva Trust Center

7. Microsoft Copilot

Microsoft Copilot

7/10

GOOD PRIVACY

Copilot offers controls and enterprise guarantees; personal identifiers are stripped for training and enterprise customers get contractual protections.

Source: Microsoft Privacy | Copilot Support

8. Anthropic Claude

Anthropic Claude

8/10

EXCELLENT PRIVACY

Anthropic doesn't use user data for training by default and emphasizes encryption and limited employee access, making it a strong choice for privacy-conscious users.

Source: Anthropic Privacy | Anthropic Support

Essential Privacy Protection Tips

Never enter private or confidential information into ChatGPT and similar tools. By default, ChatGPT and other AI tools may use your data to improve performance. Always assume your inputs could be used for model improvement unless explicitly opted out.

Use temporary or ephemeral chat modes for sensitive discussions. Pay for enterprise versions when you need contractual data guarantees. Most importantly, read privacy settings and change defaults - many protective controls exist but remain disabled by default.

Your AI conversations aren't as private as you think, but knowing these rankings helps you choose wisely. For maximum privacy, stick with Claude or use self-hosted solutions when handling sensitive information.