Charles Hoskinson, founder of Cardano, released Anthropies, a free tool designed to strip watermarks that Anthropic embeds in Claude-generated text, code, and images. The tool addresses growing concerns around AI attribution and content provenance in the crypto and broader tech communities.
Anthropic, the AI safety company behind Claude, embeds watermarks in outputs to track and authenticate content generated by its models. These watermarks serve as a form of digital fingerprinting, helping identify AI-generated material and maintain transparency about content sources. Hoskinson's release of a watermark removal tool challenges this approach, positioning it as a freedom-of-information initiative rather than an anti-attribution mechanism.
The move reflects ongoing tensions in the AI ecosystem around watermarking, model attribution, and user rights over generated content. Crypto communities have historically pushed back against centralized control mechanisms and mandatory tracking systems, viewing them as restrictions on user autonomy. Hoskinson frames Anthropies within this ideological context, offering the tool as an open alternative for users who want to use Claude outputs without embedded metadata.
Anthropic has not publicly responded to the tool's release. The company's watermarking strategy aligns with broader industry efforts to combat AI-generated spam, deepfakes, and unauthorized content replication. However, the existence of removal tools raises questions about the practical effectiveness of such safeguards.
Hoskinson's involvement connects the tool to Cardano's broader ecosystem, though Anthropies operates independently of the blockchain platform. The release underscores how crypto thought leaders engage with AI governance issues, often favoring decentralized or permissionless alternatives to corporate-controlled systems.
The tool's availability on a free, public basis signals potential wider adoption among users skeptical of content attribution requirements. Whether Anthropic pursues legal action or technical countermeasures remains unclear, but the incident highlights the ongoing friction between AI companies prioritizing model provenance and communities prioritizing user freedom.