The artificial intelligence (AI) startup noted the change on the support page for its Claude model Tuesday (Aug. 11), saying it came in response to the European Union’s AI Act’s Code of Practice on Transparency of AI-Generated Content.
“As AI-generated content becomes commonplace, greater transparency and signals about where content comes from can give people useful context about the information they consume,” Anthropic wrote. “To support transparency and comply with our legal obligations, Anthropic is working to include machine-readable marks in content that Claude generates.”
According to the announcement, Claude models launched in the EU on or after Aug. 2—the day the new regulations went into effect—will support machine-readable marking at launch, with watermarks embedded into AI-generated text. Generated files will include “digitally signed provenance metadata where supported,” the company added.
The marking works everywhere people encounter Claude, Anthropic said, applying to output from supported Claude models across the company’s stable of models. The embedded watermarks also apply when users access Claude via Anthropic’s cloud partners at Amazon, Google and Microsoft.
PYMNTS wrote about the new EU rules last week, noting that they arrive as California is preparing to institute its AI Transparency Act next year.
While the California legislation is somewhat more narrowly focused, the report said both laws arrive at a similar conclusion: organizations need to go beyond just informing users when AI has been used and develop systems to allow them to detect AI-created content.
“For businesses, particularly regulated financial institutions, the shift signals the emergence of a new enterprise AI governance challenge,” that report said. “Rather than treating AI transparency as a consumer disclosure obligation, companies are increasingly being required to embed technical controls, governance processes and audit capabilities into the way AI-generated content is created, distributed and managed.”
The implications are especially important for financial institutions, which are increasing their use of generative AI in areas such as customer service, fraud detection, document preparation and internal operations.
“Banks already operate under extensive supervisory expectations governing model risk management, cybersecurity, operational resilience and third-party risk management,” the report continued. “AI transparency obligations increasingly intersect with each of those existing governance frameworks.”
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
