MOSTLY AI

?? ?????????? ???????????? ?????????? ?? ?? ?????? ?????? ???? ???? & ???????? ???????????????????????????????? At MOSTLY AI, our mission has always been to democratize data. To empower every business, every entity to harness the full potential of their proprietary data without risking #privacy. With this MAJOR announcement, we aim to remove the remaining friction towards adopting #syntheticdata at scale, while boosting transparency and trust. We proudly present the first industry-grade open source SDK for privacy-safe synthetic data — developed to meet and exceed the expectations of modern, large-scale organizations: ??state-of-the-art accuracy ??state-of-the-art privacy (incl. differential privacy) ??best-in-class compute efficiency (16x to 100x faster) ??deploy in your secure compute environment ??broad range of supported data types ??fully permissive Apache 2.0 license ?? Ready to explore? Join our community by installing, using, and building upon our ?????????????????? ???????? ?????? (???????? ??????????). Share it with your data teams and shape the future of AI-driven innovation! #datademocratization #artificialintelligence #opensource #gosynthetic

Will the U.S depriorization of responsible and trustworthy AI backfire in the long-term? ?? I spoke on Schwab Network about why reaching a sustainable balance between innovation and regulation now will help society benefit from trustworthy AI in the future. Thanks so much for having me on the show Jenny Horne and Alex Coffey – I really enjoyed our chat! And thanks also for your help in setting up the opportunity, Kaitlyn Crist. Some of the key talking points: → The importance of regulation and guardrails that don't hinder innovation → Potential long-term impacts of focusing on innovation over responsibility and trust → What progress in car safety mechanisms tells us about how AI trust and safety will develop over time → How data democratization will enable the next generation to get more value from AI – including personalized healthcare and financial advice I hope the overarching message was that I'm optimistic for trustworthy AI that is a net positive for society in the mid-term. The scrambling to get it right we’re seeing now is to be expected –?but I’m confident we’ll get there! ?? Here's a clip if you're keen to learn more. ?? #responsibleAI #syntheticdata #datademocratization #AIethics

?? NEW FEATURES ?? ??? Only two weeks ago, we introduced our open-source Synthetic Data SDK—a cutting-edge Python package for generating differentially private synthetic data right in your local environment, giving you full control over privacy. ?? Today, we’re excited to launch MOSTLY AI Platform v4.1.0 ?? Delivering data without boundaries that lets you import locally generated synthetic data assets into the platform for collaboration and sharing. This empowers you to move from solo experimentation to team-wide or even public use—while maintaining privacy at every step. ?? Organizations ?? ? Create an organization, invite your team, and manage access. ? Centralize your synthetic data assets in one place for streamlined collaboration. ?? Private or Public Assets ?? ? Keep datasets exclusive to specific users or share them publicly. ? You decide how far your synthetic data travels. ?? We’d love your feedback! Try out the latest features and let us know what you think!

"???? ?????? ???????? ?????????? ???? ???????????? ?????? ??????????, ???????????? ????????????, ?????? ?????????????? ???????????? ?????? ?????????????? ???????????? ?????? ??????. ???? ??????'???? ?????? ?????????? ???? ???????? ???? ???????? ???? ?????? ?????????????? ????????????, ????????, ?????? ??????????????????????, ?????? ???????????????????? ???????? ?????? ???????????? ?? ??????????????.?" That’s why we built the first industrial-grade open source #syntheticdata SDK — because both enterprizes and society at large need high-quality, privacy-safe data to drive meaningful AI progress. Cate Lawrence just wrote what’s easily my favorite piece on synthetic data and MOSTLY AI's mission of #datademocratization. Cate, thank you so much for taking the time to interview me and publishing this incredible piece on synthetic data. You did an incredible job capturing why this matters — not just for enterprizes, but for #AIInnovation at large. https://lnkd.in/dcZXkPeq #AIActionSummit #Paris2025 #AI #Privacy #ArtificialIntelligence #AIAct #OpenSource #ResponsibleAI #OpenData #OECD Tech.eu #AIOpportunitiesActionPlan Peter Kyle OECD.AI Dr. Clara Neppel Alpesh S. IEEE European Data Protection Board Jiri Hradec Malte Beyer-Katzenberger European Commission Axel Voss Kai Zenner Cameron Kerry Andrea Renda CEPS (Centre for European Policy Studies) Bojana Bellamy Hunton & Williams LLP National Institute of Standards and Technology (NIST) EDPS - European Data Protection Supervisor The Royal Society

?? Unlock data democratization and accelerate model development on Amazon SageMaker with privacy-preserving Synthetic Data from MOSTLY AI ?? Check out our latest blog, co-authored by Julio Zambrano and Siva S. ?? https://lnkd.in/dXWZ8JXf We're excited about deeper integration with SageMaker Unified Studio in 2025! ?? Faris Haddad Thomas Zerbach Venkatesh Krishnan, PhD Alexandra Ebert Mario Scriminaci John Sullivan #mostlyai #aws #generativeai #syntheticdata #datascience

?? Now supporting Databricks compute!??? You can now use your Databricks clusters for generator training and synthetic data generation, alongside Kubernetes. This update unlocks new possibilities, providing more flexibility for your workflows. #syntheticdata?#databricks?#ai #generativeai #mostlyai #datascience

?? Need mock data for your next project? Our latest Product Tips video is here to help! Watch as our CPO, Mario Scriminaci, demonstrates how to generate mock data effortlessly using MOSTLY AI. ?? And here’s the kicker: stick around until the end of the video for a bonus tip! If you need larger datasets, Mario shares how to scale up with ease. ??? Watch now and supercharge your mock data generation: https://lnkd.in/d7RDRZjd #SyntheticData #MockData #AI #DataGeneration #ProductTips

??? From Table Talk to Data Magic! In our last video, we introduced the basics of setting up table relationships in a simple two-table scenario. But what if your data structure is more complex? In this follow-up video, Zahari Tilev dives into multi-table relationships, explaining how to use context vs. non-context foreign keys to ensure your synthetic data is accurate and meaningful. ?? Watch it here: https://lnkd.in/dTuSxg5W We’d love to hear your thoughts—share your feedback or questions in the comments! #SyntheticData #TableRelationships #DataPrivacy #GenerativeAI?#GenAI #MostlyAI

?? Exciting updates inside! At MOSTLY AI, we’re always working to improve your experience with synthetic data. Here's what's new: ??? Differential Privacy You now have the choice to train generators with or without differential privacy (DP) guarantees! ? Training becomes differentially private, with the platform tracking the privacy budget (Epsilon) throughout. ? Set an upper Epsilon limit to automatically stop training when the budget is reached. Want to learn more about how differential privacy works and why it matters? Check out our blog post: ?? https://lnkd.in/d6_PzXmE ?? Notification Center No more wondering about training and generation progress! The Notification Center keeps you informed with: ? Updates on the success and failure of Generator training and Synthetic dataset generation. ? Notifications when a peer shares a Generator or Synthetic dataset with you. ?? Likes for Generators and Synthetic Datasets Spread the love! Whether it’s a generator or synthetic dataset you admire, click the ?? to show your appreciation and share feedback with creators. Explore the possibilities of synthetic data today—join us at app.mostly.ai to get started. Which feature are you most excited about? Let us know in the comments!

There are two camps within privacy engineering: 1) the Formalists ????, who consider theoretical guarantees above all 2) the Pragmatists ????, who care for practical & effective solutions that work Happy to share that MOSTLY AI now proudly serves both camps with ???????????????????????????? ?????????????? ?????????????????? ???????? - available for TABULAR as well as for LANGUAGE models ?? Learn more in our latest blog post: https://lnkd.in/dsCS3V_Z