Anthropic's AI Watermark: Claude Users Revolt

Anthropic is facing a backlash from some Claude users over its new watermarking system for AI-generated text. Many users are reportedly canceling their subscriptions, with some posting their frustrations online. Here's the thing: Anthropic announced this watermarking effort to comply with the European Union’s AI Act. Influencers and other users are citing this "ridiculous watermark idea" as the reason for ending their service. What's interesting is how the watermarking works. Anthropic says the difference between watermarked and un-watermarked text will not be distinguishable to readers. It uses a system similar to Google's SynthID, where token choices are subtly steered by a secret key, especially in situations where multiple words could fit. This is not about crude symbol replacements. The bottom line: This system aims to make it harder to pass off AI-generated text as human-written, but it's clearly upsetting a segment of its user base.