Why We're Making These Changes
We believe in providing a private, secure messaging platform for everyone. Unfortunately, a small minority of users have abused our services to send spam, phishing attempts, and other harmful content. This has led to situations where we've had to suspend accounts and, in some cases, faced pressure from carriers and regulators.
We're genuinely sorry. The actions of a few bad actors create problems for the entire community. We'd much rather focus on building great features than implementing restrictions, but protecting our platform, and you, requires us to act.
Introducing Smart Message Filtering
Starting today, all outgoing SMS messages are processed through an LLM-powered content filter before being sent. Here's what you need to know:
How It Works
- You compose your message as normal
- The message enters a brief queue (typically under 5 seconds)
- Our LLM analyzes the content for spam, phishing, fraud, and other harmful patterns
- If approved, the message is sent and credits are deducted
- If blocked, no credits are charged and you'll see the message marked in red
Privacy-First Design
We take your privacy seriously. Here's how we've built this system with privacy at its core:
- 🔒 No message storage - We do not store your messages. The filter processes content in real-time and immediately discards it.
- 🔐 End-to-end encryption - All communications remain encrypted internally throughout the entire process.
- 🤖 Self-hosted LLM - We run our own instance of Gemma 3, an open-source model from Google, with our custom safety instructions and fine-tuning.
- 🏠 On-premises processing - Nothing leaves our infrastructure. No third-party APIs, no cloud LLM services, no data sharing.
The LLM model only sees the message content momentarily to classify it, it has no access to your identity, phone numbers, or any other personal information. Once classification is complete, the content is gone.
What Gets Blocked?
Our filter targets:
- 📵 Spam and bulk marketing
- 🎣 Phishing attempts
- 💳 Financial fraud and scams
- 🔗 Malicious links
- ⚠️ Other harmful content
Legitimate messages are not affected. Normal conversations, verification codes, business communications, and personal messages will pass through without issue.
Blocked Messages in Your Conversation
If a message is blocked, you'll see it appear in your conversation with a red indicator and "Blocked by content filter" status. You are not charged for blocked messages.
New Rate Limits
To further protect against abuse, we've introduced sensible rate limits on outgoing messages:
Platform (Web & App)
| Timeframe | Limit |
|---|---|
| Per Minute | 5 messages |
| Per Hour | 60 messages |
| Per Day | 300 messages |
API
| Timeframe | Limit |
|---|---|
| Per Minute | 10 messages |
| Per Hour | 100 messages |
| Per Day | 500 messages |
These limits are per-user and reset on a rolling basis. For the vast majority of users, you'll never hit these limits during normal use.
Need higher limits? If you have a legitimate business use case requiring higher volumes, please contact our support team to discuss your needs.
What If My Message Was Incorrectly Blocked?
We understand that no automated system is perfect. If you believe your message was incorrectly blocked:
- Open a support ticket through your account dashboard
- Select "Technical" as the category
- Describe the message you were trying to send and why you believe it was incorrectly flagged
Our team will review your case and, if appropriate, whitelist similar content patterns for your account.
Our Commitment to You
These measures aren't about restricting legitimate users, they're about ensuring Crypton remains a trusted, reliable platform for everyone. We're committed to:
- ✅ Transparency - You'll always know when a message is blocked
- ✅ Fairness - No charges for blocked messages
- ✅ Privacy - Zero message storage, fully encrypted, self-hosted open-source LLM
- ✅ Improvement - We continuously refine our filters to reduce false positives
Questions?
If you have any questions about these changes, please don't hesitate to reach out through our support system. We're here to help.
Thank you for being part of the Crypton community. Together, we can keep this platform safe and useful for everyone.
- The Crypton Team
