We take a value-driven approach to building AI-enabled communication technology. We leverage AI and other technologies only to address actual challenges people face in communicating their ideas and being understood as intended.

We build products and models with checks and balances to prioritize privacy, safety, and fairness. We rigorously evaluate our work to anticipate its impact on our users and communities.

People trust us with their words, and we earn their trust by putting data security and privacy at the core of our business and our product. With 15 years of experience developing best-in-class AI communication assistance, we will always go to great lengths to ensure user data is encrypted, private, and secure. Users can control whether their content is used to improve our models, and we do not allow any partners or third parties to use our customers’ content to train their models or improve their products.

We put users in control of their experience. AI is a tool that helps augment communication, but it can’t do everything. People are the ultimate decision-makers and experts in their own relationships and areas of expertise. Our commitment is to help every user express themselves in the most effective way possible.

When you use Grammarly’s products, you’re trusting us to handle your personal information with care. This means that we do not and will not sell your data. We make money by selling subscriptions to our services.

Using a combination of technologies, we filter generative AI and natural language suggestions to address issues, such as hate speech, should they arise. Our integrations and models help generate more effective text and reduce risks in input and output.

We are committed to building models using quality data sets that undergo bias and fairness evaluations. We design and develop products with our team of analytical linguists who apply research and expertise to minimize bias and apply user feedback.

Every new feature goes through a rigorous risk-assessment process, including hands-on linguists reviewing output to identify potential risks. Once any risks are assessed, feature teams are required to make updates.