Tokens are the basic units that our AI system uses to process text. Think of tokens as small pieces of text that the AI reads and understands. A token can be a word, part of a word, or even a single character depending on how common or complex it is.
For example:
- Simple words like "the" or "and" are usually one token each
- Longer words like "complicated" might be broken into multiple tokens
- Special characters, numbers, and spaces also count as tokens
Tokens are important because they determine how much text you can send to and receive from the AI in a single conversation. Your subscription plan includes a specific number of tokens that you can use each month.
Tokens are calculated based on both the text you send to the AI (your questions or content) and the text the AI generates in response. Here's how it works:
- Input tokens: These are counted from the text you provide to the AI, including your questions, uploaded documents, or any content you want the AI to analyze.
- Output tokens: These are counted from the AI's responses to your queries.
- Total usage: Your token usage is the sum of both input and output tokens across all your conversations.
Different types of content use different amounts of tokens:
- A page of text (about 250 words) typically uses around 350-500 tokens
- Complex content with technical terms or unusual words may use more tokens
- Simple, common phrases use fewer tokens
You can monitor your token usage in your account dashboard to see how many tokens you've used and how many remain in your current billing period.
Your token allocation automatically renews every 30 days from the start of your subscription period. Here's how the token renewal system works:
- Regular renewal: Your tokens automatically refill to your plan's maximum amount every 30 days.
- Upgrade options: If you need more tokens before your next renewal date, you have two options:
- Upgrade to a higher tier plan with more tokens
- Purchase additional token packs to supplement your current plan
- Rollover policy: Unused tokens from your previous period do not roll over to the next period. When your tokens refill, they reset to your plan's allocation amount.
- Usage notifications: You'll receive notifications when you're approaching your token limit so you can decide whether to conserve usage or purchase more tokens.
If you frequently run out of tokens before your renewal date, consider upgrading to a plan that better fits your usage patterns.
We take the privacy of your content extremely seriously, especially since we understand authors are concerned about protecting their creative work. Here's how we safeguard your data:
- No training on your content: Unlike some AI services, we do not use your input or output text to train or improve our AI models. Your creative work remains yours alone.
- Data isolation: Your conversations and uploaded content are stored securely and are not shared with other users or third parties.
- Limited data retention: We only retain your data for as long as necessary to provide our services to you, and you can request deletion of your data at any time.
- Encryption: All data transmitted between your device and our servers is encrypted using industry-standard protocols.
- Access controls: We implement strict access controls within our organization, meaning only authorized personnel can access user data when absolutely necessary for support or troubleshooting.
- Transparency: We're transparent about our data practices and will notify you of any changes to our privacy policies.
Your creative work is valuable, and we've designed our system specifically to address the concerns of authors and content creators who want to use AI tools without risking their intellectual property.
This is an important question that many authors have. Here's what you should know about Anthropic's Claude and copyright:
- Training data sources: Like all large language models, Claude was trained on a diverse dataset that includes publicly available text from the internet, books, and other sources. This training data likely included some copyrighted materials, as is standard in the AI industry.
- Legal compliance: Anthropic has stated they follow legal frameworks around fair use and copyright in their training process. They've designed their systems to respect intellectual property rights.
- No verbatim reproduction: Claude is designed to generate original text rather than reproduce copyrighted content verbatim. The AI synthesizes information rather than copying exact passages from its training data.
- Continuous improvement: Anthropic regularly updates their models and policies to better address copyright concerns and improve how they handle intellectual property issues.
- Your content remains yours: Most importantly, using our service doesn't grant Anthropic or us any rights to your creative work. The content you create with our tool belongs to you.
It's worth noting that the legal landscape around AI and copyright is still evolving. We stay informed about developments in this area and are committed to maintaining ethical practices that respect authors' rights while providing valuable AI assistance.
If you have specific concerns about copyright protection for your work, we're happy to discuss additional safeguards or answer any questions you might have.