Grammarly Earns ISO/IEC 42001:2023 Certification

At Grammarly, we’re dedicated to constructing AI that helps efficient, considerate communication—with out compromising on security, privateness, or person belief. That dedication has all the time guided how we design, develop, and deploy our AI-powered writing help. Now, it’s backed by an internationally acknowledged customary: Grammarly is formally ISO/IEC 42001:2023 licensed.

As AI turns into more and more central to the way in which we work, write, and join, the necessity for clear tips round its moral growth and use has by no means been better. Enter ISO 42001—the world’s first worldwide customary targeted particularly on managing AI techniques responsibly.

What ISO/IEC 42001:2023 is—and why it issues

ISO 42001 helps organizations set up a framework for governing AI in a approach that’s safe, clear, and aligned with human values. This certification validates what Grammarly has prioritized from the start: constructing AI that enhances communication whereas proactively assessing threat and safeguarding person knowledge. ISO 42001 compliance reveals that Grammarly’s AI is:

• Ethically designed to align with human values
• Aligned with finest practices and an eye fixed towards rising laws
• Safe and privacy-conscious to guard delicate knowledge
• Repeatedly monitored for dangers and unintended penalties
• Reliable and dependable for enterprise and person functions

Grammarly’s certification means our strategy to AI has been independently verified in opposition to this sturdy framework—making us one of many first firms globally to realize this milestone.

How Grammarly aligns with ISO 42001

We’ve embedded accountable AI rules into each a part of our product growth lifecycle. Right here’s how our strategy maps to ISO 42001’s key management areas:

Insurance policies that prioritize duty

Grammarly maintains a proper set of AI insurance policies that information the event, deployment, and oversight of our AI techniques. These insurance policies are reviewed yearly and replicate our dedication to security, privateness, and user-centric innovation. We’ve established an Synthetic Intelligence Administration System (AIMS) and conduct unbiased audits to make sure accountability.

A devoted accountable AI group

We’ve constructed a cross-functional group that oversees AI governance and ensures alignment with moral rules. Roles and obligations are clearly outlined inside our organizational construction. All staff obtain ongoing coaching in AI ethics, privateness, and safety practices.

Rigorously managed AI sources

Our datasets and third-party instruments are rigorously reviewed and secured. We preserve strict entry controls and prohibit companions from utilizing any buyer content material to coach their fashions. Our generative AI is powered by trusted companions like Microsoft Azure OpenAI and OpenAI, however buyer knowledge isn’t shared for mannequin coaching.

Proactive threat and affect assessments

We conduct thorough AI-specific threat assessments, together with pink teaming and adversarial testing, to judge equity, security, and bias. Customers are empowered to report considerations, and we repeatedly monitor AI outputs by way of automated and human analysis.

A full lifecycle strategy to AI

Our AI techniques are ruled at each stage—from design and growth to deployment and decommissioning. Earlier than releasing a brand new function, we run intensive testing to make sure it meets excessive requirements for high quality, inclusivity, and usefulness.

Safe, truthful, and clear knowledge use

Grammarly enforces robust governance over the info used to construct our AI, emphasizing high quality, equity, and privateness. Customers can management how their content material is used, and we by no means promote person content material.

Clear, accessible info for stakeholders

Transparency is a core precept at Grammarly. We offer explanations for AI-generated options and preserve clear documentation round our privateness and safety practices in our Consumer Belief Heart. Devoted buyer help and success groups can be found to reply AI-related questions.

Human-centered AI use

Grammarly’s AI is designed to help—not change—human creativity and judgment. Customers all the time stay answerable for how they interact with AI options, and our techniques are constructed with transparency and person company on the forefront.

Accountable third-party partnerships

We maintain our distributors and subprocessors to the identical excessive requirements we set for ourselves. All companions endure annual safety and compliance evaluations, and we’re clear about the place third-party AI fashions are used. Grammarly holds enterprise-grade certifications together with SOC 2 (Sort 2), ISO 27001, and ISO 27701, and is compliant with HIPAA, CCPA, GDPR, and extra.

Constructing AI you’ll be able to belief

Reaching ISO/IEC 42001:2023 certification is greater than a milestone—it’s a mirrored image of how deeply we care about constructing AI responsibly. Our accountable AI group ensures that security, equity, and transparency are embedded into each mannequin we create and each person expertise we ship.

We’re dedicated to serving to people and companies confidently use AI to speak extra successfully—with out compromising on belief.

Be taught extra about our strategy at grammarly.com/responsible-ai, and take a look at our white paper to discover our accountable AI practices in depth.