Privacy & security in MOSTLY AI's Platform
Privacy-by-design and maximum security for each and every one of your synthetic datasets.
Get started free
No 1:1 relationship to
the original data
In contrast to traditional anonymization techniques, MOSTLY AI uses your original data only as learning material to train Generative AI models. During training, the models learn the patterns, distributions, correlations, and other statistical characteristics of your original data. MOSTLY AI then uses the AI models to generate synthetic data from scratch. Thus a synthetic record cannot be linked back to one specific original data record. Instead it is based upon the input of what was generalized by the model from all original data records.
Model overfitting
prevention
We've implemented a robust mechanism to prevent our Generative AI models from memorizing individual properties and patterns. Our approach involves carefully designed loss functions and validation criteria, all aimed at ensuring generalization and guarding against overfitting. The models only learn general patterns of the original data, but no specific individual data points.
Read more about model overfitting prevention in our DocumentationRandom draw
synthesis
To generate synthetic data, the MOSTLY AI Platform generates new samples with random draws against the trained AI models. Let's consider a simplified example with a column like 'Gender,' which has categories like Male, Female, Other, and N/A. The model learns the distribution, like 47% females, 45% males, 7% other, and 1% N/As. During a draw, the chance of 'Male' is about 4-5 times out of 10.
As mentioned, this example is overly simplified as during each random draw, the MOSTLY AI Platform considers not only the distribution of a single column but also all statistical characteristics and the relationships between each column of the original data.
Privacy protection mechanisms
Our commitment to privacy extends to safeguarding against re-identification risks, especially in scenarios involving rare categories, extreme values, and extended sequence lengths.
Learn more in the MOSTLY AI DocumentationRare category protection
The Platform uses rare category protection for categorical columns, preventing the AI model from being trained with rare values. To maintain the original data's correlation and distribution, we substitute these values with the category "_RARE_".
Extreme value protection
The Platform applies extreme value protection to numerical and date-time columns. Before training, it removes minimum and maximum outliers from these columns to prevent exceptional cases from appearing in the synthetic data.
Extreme sequence length protection
The Platform removes excessive linked records that could lead back to a subject in a subject table. Long sequences can pose a privacy risk, so they are removed before training of the Generative AI model.
Privacy settings by default
In the MOSTLY AI Platform, all configuration settings to protect data privacy are on by default, so you can rest assured.
Why should you trust MOSTLY AI's synthetic data?
SOC 2 and ISO 27001 certified solution with maximum security
Continuous external audits and legal assessments for compliance
The highest data anonymization standards
Complies with the requirements of GDPR, CCPA, CPRA, HIPAA, PDPA, & APPI
Available for on-premises installations, including in air-gapped environments, or deployed in private cloud infrastructures
Trusted by leading brands and data privacy experts
GDPR-compliant data anonymization by default
MOSTLY AI's Platform provides complete anonymization by default. Preconfigured settings for non-expert users eliminate human error. Automatic, state-of-the-art privacy mechanisms ensure that your data is safe and your customers will be protected against threats.
No risk of data breaches & data fines
With MOSTLY AI's synthetic data it is not possible to single out an individual, link records relating to an individual, or infer information concerning an individual.
Your customer data is kept safe and secure. Synthetic data helps eliminate the risk of data breaches and privacy fines.
Ready to try synthetic data?
The best way to learn about synthetic data is to experiment with synthetic data generation. Try it for free or get in touch with our sales team for a demo.