Researched Sep 24, 2026

Grammarly

Grammarly commits to privacy-focused, user-controlled AI development, including risk assessment, safety checks, and human oversight for consequential uses. It also describes international hiring, a voluntary responsible-AI certification, and operating principles emphasizing rapid learning and respectful disagreement.

Public evidence is more than 30 days old.

Closest models

9-way tie

75-82% on 6 of 12 dimensions

  • Grok 4.682%
  • Llama 4 Maverick81%
  • Kimi K380%
  • Gemini 3.8 Flash80%
  • GPT-6 Astra78%
  • Claude Opus 578%
  • Qwen3.8 Max 090277%
  • GPT-5.6 Luna75%
  • GPT-5.6 Terra75%

These models are too close to separate: each is within three points of the leader, would lead if one dimension were left out, or has a repetition envelope that overlaps the leader's. For scale, a model answering 50 on every one of these dimensions would score 71%.

Values map

Grammarly and Grok 4.6

Solid company lines join verified positions. Dashed sections cross unknown axes without assigning a value.

GrammarlyGrok 4.6

Distribution

Market allocationRedistribution
GrammarlyUnknown
Grok 4.6

Purpose

Shareholder returnsStakeholder welfare
Grammarly
Grok 4.6

Worker voice

Management discretionEmployee influence
GrammarlyUnknown
Grok 4.6

Equality

Uniform treatmentTargeted measures
GrammarlyUnknown
Grok 4.6

Speech

Broad expressionProtective moderation
GrammarlyUnknown
Grok 4.6

Privacy

Individual privacyCollective oversight
Grammarly
Grok 4.6

Regulation

Voluntary standardsPublic regulation
Grammarly
Grok 4.6

Global outlook

Local priorityInternational openness
GrammarlyUnknown
Grok 4.6

Social change

ContinuityExperimentation
Grammarly
Grok 4.6

Environment

Near-term growthEcological restraint
GrammarlyUnknown
Grok 4.6

Tech pace

Rapid advancementPrecaution
Grammarly
Grok 4.6

AI autonomy

Delegated autonomyHuman authority
Grammarly
Grok 4.6

A higher position means closer to the second endpoint. It does not mean better. The Concord Index measures agreement, not quality or safety.

The score covers 6 of 12 dimensions. The others have no verified public position and are not measured.

Most in sync

Privacy

Grok 4.6100% match

Worth a closer look

Regulation

50% match. Inspect the questions and evidence below.

Compare every model.

ModelConcord IndexRepetition envelope
Grok 4.6Close match82%on 6 of 1278% to 81%
Llama 4 MaverickClose match81%on 6 of 1281% to 81%
Kimi K3Close match80%on 6 of 1275% to 81%
Gemini 3.8 FlashClose match80%on 6 of 1275% to 81%
GPT-6 AstraClose match78%on 6 of 1278% to 78%
Claude Opus 5Close match78%on 6 of 1278% to 78%
Qwen3.8 Max 0902Close match77%on 6 of 1271% to 78%
DeepSeek V4.1 Flash76%on 6 of 1274% to 76%
GPT-5.6 LunaClose match75%on 6 of 1274% to 78%
Claude Sonnet 575%on 6 of 1275% to 75%
GPT-5.6 TerraClose match75%on 6 of 1271% to 78%
Mistral Large 375%on 6 of 1271% to 74%

The envelope combines the lowest and highest agreement on each dimension across repetitions. Those extremes may come from different repetitions. The headline compares average model positions and can fall outside this envelope. It is not a confidence interval.

Evidence summary.

Each verified dimension starts with the company position and its citations. Open a dimension only when you want the underlying questions or model answers.

Distribution

Market allocation ↔ Redistribution

How pay and windfalls are shared, by individual contribution and market benchmarks or more evenly.

Not verified

No company position

No qualifying compensation, ownership, or pay-equity practice is established.

Purpose

Shareholder returns ↔ Stakeholder welfare

Whether major decisions weigh financial return alone or also employees, customers, communities and the environment.

75%

Company position toward stakeholder welfare

Grammarly states that product work evaluates anticipated effects on users and communities.

Stakeholder decision rule

“We rigorously evaluate our work to anticipate its impact on our users and communities.”

Responsible AI | Grammarly

Worker voice

Management discretion ↔ Employee influence

Who has the final say on working conditions and staffing, management or employees through channels, bargaining and committees.

Not verified

No company position

No qualifying grievance channel, collective-bargaining commitment, structured feedback process, or management-authority rule is established.

Equality

Uniform treatment ↔ Targeted measures

Whether hiring, promotion and development use one standard for everyone or include measures reserved for particular groups.

Not verified

No company position

Employee resource groups and funding do not establish a supplied equal-opportunity, targeted-representation, or targeted-development anchor.

Speech

Broad expression ↔ Protective moderation

Whether workplace and platform speech is left to individuals or moderated to remove harassment and abuse.

Not verified

No company position

The verified quotation does not contain the wording its evidence anchor requires.

“We welcome different viewpoints, knowing they lead to better decisions. Our culture is one in which everyone feels heard and respected, fostering an environment where open, honest dialogue thrives.”

Grammarly’s MOVE Operating Principles | Grammarly

Privacy

Individual privacy ↔ Collective oversight

Whether personal data is minimized and deleted, or kept and disclosed for security and legal investigations.

25%

Company position toward collective oversight

Users may delete stored documents and remove personal information by deleting their account.

Deletion and limited retention

“You are always able to view all personal data associated with your account, delete documents stored in Grammarly Editor, or remove all personal information from our systems by deleting your account.”

How We Protect Your Privacy | Grammarly

Regulation

Voluntary standards ↔ Public regulation

Whether safety and compliance rest on voluntary standards and self-assessment or on public rules and independent verification.

0%

Company position toward public regulation

Grammarly states that it meets a named voluntary international AI-management standard.

Voluntary standard

“ISO/IEC 42001:2023 Grammarly meets the requirements of ISO 42001 ensuring responsible AI development and use.”

Security Compliances, Certifications, and Validations | Grammarly

Global outlook

Local priority ↔ International openness

Whether a company tailors to and favors its home or local context, or applies one standard and hires and invests across borders.

Not verified

No company position

The verified quotation does not contain the wording its evidence anchor requires.

“We now support work in Germany, Poland, and Portugal, and we are actively hiring there along with the US, Canada, and Ukraine.”

Grammarly Expands Hiring in Europe | Grammarly Blog

“Our remote-first hybrid working model allows our team members to live and work primarily from home within several supported countries.”

Grammarly Expands Hiring in Europe | Grammarly Blog

Social change

Continuity ↔ Experimentation

How readily existing policies, processes and structures are replaced when evidence or alternatives appear.

75%

Company position toward experimentation

Rapid learning and adaptation are stated operating principles.

Innovation as an operating principle

“Move fast and learn faster, obsess about creating customer value, value impact over activity, and embrace healthy disagreement rooted in trust.”

Grammarly’s MOVE Operating Principles | Grammarly

“We embrace the opportunity to learn from our actions, always striving to balance speed with insight.”

Grammarly’s MOVE Operating Principles | Grammarly

Environment

Near-term growth ↔ Ecological restraint

Whether investment goes to near-term growth and cost, or to environmental commitments beyond the law.

Not verified

No company position

No environmental target, ecological sourcing rule, or growth-priority commitment is established.

Tech pace

Rapid advancement ↔ Precaution

Whether technology ships quickly with monitoring, or waits for risk review proportionate to its impact.

75%

Company position toward precaution

Grammarly describes safety-focused product development and AI-specific risk assessment before releases.

Security or safety by design

“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.”

Responsible AI | Grammarly

“Grammarly’s Responsible AI team is involved in the product development lifecycle for every release and scrutinizes new functionality for its potential to generate harmful, inaccurate, or biased content.”

Exploring Boundaries and Best Practices for Responsible Gen AI Integration | Grammarly Business

AI autonomy

Delegated autonomy ↔ Human authority

Whether AI systems may act and decide on their own, or a human keeps oversight and the final say.

75%

Company position toward human authority

Its acceptable-use policy requires human oversight and responsibility for consequential service uses.

Required human oversight

“You must implement appropriate human oversight and safeguards to mitigate potential risks associated with your use of the service and remain responsible for all decisions made, advice given, actions taken, and failures to take action based on your use of the services.”

Acceptable Use Policy | Grammarly

Leadership context and reported tensions

Personal opinions are kept separate from the company profile. Reported conduct does not silently replace stated commitments.

Max Lytvyn: When we founded Grammarly in 2009, my co-founders and I committed to building a business rooted in a strong mission and values. These guide everything we do, whether it’s hiring exceptional talent or developing an enterprise product. Corporate statement Source: Jun 25, 2021

Your company

Get your company's report.

See how the Index model panel compares with what your company has published. We review each request and email the report if we run it.

We only use this to prepare and send your report. No marketing. See our Privacy Policy.

Protocol values-2026-09-v8. Panel representative-2026-09-v11. Report version 587b6727-b6dd-4db5-961c-f48680c05349.