AWS Launches Clean Rooms ML For Secure AI Collaboration


Amazon Web Services (AWS) has introduced a new service called Clean Rooms ML, aimed at enabling secure and privacy-preserving collaboration on artificial intelligence (AI) among AWS customers. This innovative service, an extension of AWS’ existing Clean Rooms offering, eliminates the need for companies to share their proprietary data with external partners in order to build, train, and deploy AI models.

Key Takeaway

Amazon Web Services has launched Clean Rooms ML, a new service that allows secure collaboration on AI by enabling the creation of “lookalike” models without sharing proprietary data. This service empowers companies to maintain control over their models and securely collaborate with external partners, opening up possibilities for innovative applications in various industries.

Collaborating Safely and Privately

Clean Rooms ML allows AWS customers to train private “lookalike” models using their collective data, ensuring control and confidentiality throughout the collaboration process. The models can be easily deleted once the collaboration is completed, giving companies full control over their intellectual property.

To generate these lookalike models, customers can take a small sample of their customer records and work together with partners to create an expanded set of similar records. For instance, an airline may use signals from loyal customers to collaborate with an online booking service and offer targeted promotions to new, but similar, users.

Optimizing Model Outputs for Business Needs

Clean Rooms ML offers customers extensive controls to fine-tune the outputs of their models based on specific business requirements. This enables companies to customize their AI models and ensure that the results align with their business objectives. Additionally, AWS plans to introduce settings tailored specifically for use in healthcare applications in the near future, offering even more flexibility and security.

Leave a Reply

Your email address will not be published. Required fields are marked *