Anthropic just rolled out a cool new feature for its Claude AI models, and it’s all about making AI responses more reliable. Called Citations, this tool lets developers link AI answers to specific sources, like emails or documents. Think of it as giving AI a way to show its work, so you know where the info is coming from.

Here’s how it works: Developers can upload source files, and the AI will automatically reference the exact sentences or passages it used to craft its answers. This is especially handy for tasks like summarizing documents, answering questions, or even handling customer support. It’s like having a built-in fact-checker to cut down on those pesky AI mistakes—or as experts call them, “hallucinations.”

Right now, Citations is only available for two of Anthropic’s models: Claude 3.5 Sonnet and Claude 3.5 Haiku. And heads up—it’s not free. Depending on the size and number of documents, there might be a small fee. For example, a 100-page document could cost around $0.30 with Sonnet or $0.08 with Haiku. Not too shabby for a feature that could save a lot of headaches.

This update comes at an interesting time, just as OpenAI is making waves with its own AI developments. Anthropic seems to be doubling down on transparency and accuracy, which could be a game-changer for businesses relying on AI for critical tasks.

So, if you’re a developer looking to make your AI tools more trustworthy, this might be worth checking out. After all, who doesn’t want fewer errors and more reliable answers?