Anthropic requires 30 day data retention for Fable and Mythos (support.claude.com)

Article summary

Anthropic is introducing a 30-day data retention policy for its Fable and Mythos models to support safety work and detect potential misuse. This policy applies to organizations using these models through various platforms, including AWS Bedrock, Google Cloud Agent Platform, and Microsoft Foundry. The retained data will be used to analyze patterns of misuse and will be deleted automatically after 30 days, unless flagged by automated trust and safety systems or required by law. Organizations can configure data retention settings through their respective platforms.

Main themes

• Data Retention
• AI Model Safety
• Cloud Security
• Compliance and Trust
• Enterprise Data Protection

What commenters say

• The 30-day data retention policy is a necessary measure to ensure the safe deployment of powerful AI models, but it raises concerns about data privacy and potential misuse.
• Some argue that the policy is unacceptable and that companies should not be allowed to retain customer data for any period, while others see it as a reasonable trade-off for the benefits of using advanced AI models.
• There is a lack of trust in companies to handle customer data responsibly, and some commenters believe that technical and legal solutions are needed to ensure data protection.
• The policy may drive some customers away, particularly those who value data privacy and security, and may lead to a loss of business for Anthropic.
• Others argue that the policy is not a significant concern, as companies already have access to customer data and that the 30-day retention period is a reasonable measure to prevent abuse.
• Some commenters question the effectiveness of the policy in preventing misuse, and whether it will actually achieve its intended goals.
• There is a debate about the balance between trust and distrust in companies, with some arguing that trust is necessary for civilization to function, while others believe that distrust is a necessary precaution in a world where companies often prioritize profits over customer interests.