01
Respect and professionalism
We engage with one another in good faith. We challenge ideas, methods and decisions without demeaning the people behind them. Disagreement is part of serious technical work; intimidation and personal attacks are not.
Sauti Labs brings people together to build and research speech and language technology. How we work matters as much as what we build.
This Code sets out the standards of conduct we expect across our collaborations, research, engineering and public work.
Our foundational standard
Respect. Integrity. Accountability.
01 / Scope
This Code applies to Sauti Labs team members, contributors, collaborators and people representing the organisation. It informs how we work together, communicate, conduct research, develop technology and participate in activities connected to Sauti Labs.
It applies in relevant professional interactions, including project discussions, meetings, research collaborations, repositories, events and public communications. Where other applicable policies or legal obligations impose additional requirements, this Code does not replace them.
02 / How we work
These principles guide our day-to-day decisions, especially when the work is difficult, the evidence is incomplete or people disagree about the way forward.
01
We engage with one another in good faith. We challenge ideas, methods and decisions without demeaning the people behind them. Disagreement is part of serious technical work; intimidation and personal attacks are not.
02
We work to make participation meaningful across different backgrounds, experiences, abilities and perspectives. We do not tolerate discrimination or harassment, and we make room for people to contribute without having to overcome avoidable barriers.
03
We communicate what we know, what we do not know and what remains uncertain. We do not knowingly misrepresent our work, capabilities, progress, evidence or commitments to teammates, collaborators, communities or the public.
04
We take responsibility for our decisions and their effects. When something goes wrong, we aim to acknowledge it, correct what we can and learn from the experience rather than conceal errors or shift blame.
05
We respect the contributions, ideas and intellectual work of others. We give appropriate credit, honour agreed responsibilities and permissions, and do not present another person's work as our own.
06
When representing Sauti Labs, we communicate within the scope of our responsibilities and avoid implying that personal views, unapproved statements or experimental results are official company positions.
03 / The work itself
Responsible technology begins with how evidence is gathered, interpreted, documented and communicated.
A
We report methods, findings, limitations and failures honestly. We do not fabricate data, falsify results, manipulate evidence or present unsupported conclusions as established findings.
B
We treat data, recordings, datasets and other research materials with care. We respect applicable permissions, licensing terms, privacy expectations and restrictions on collection, access, use and sharing.
C
We approach African languages and the people who use them with respect. We avoid treating communities as interchangeable data sources, recognise linguistic and cultural context, and take concerns about representation and harm seriously.
D
We distinguish measured performance from aspiration. We do not knowingly misrepresent a model's capabilities, evaluation conditions, language coverage, limitations or readiness for real-world use.
04 / Boundaries
Conduct that violates the standards in this Code can damage trust, compromise research and prevent people from participating safely and meaningfully.
01
Unwelcome harassment, discriminatory conduct, threats, targeted humiliation or abusive behaviour directed at people because of who they are or how they participate.
02
Misusing access, authority, confidential information or a position of responsibility to exploit, intimidate, manipulate or unfairly disadvantage others.
03
Deliberate fabrication or falsification, plagiarism, unauthorised disclosure of protected information, or knowingly misleading others about evidence, authorship or results.
04
Punishing, threatening or targeting someone for raising a concern in good faith, or engaging in conduct that undermines others' safety, dignity or ability to contribute.
Context matters. Good-faith mistakes, unsuccessful experiments, respectful disagreement and questions raised during learning are not automatically misconduct. We distinguish ordinary challenges in technical work from deliberate dishonesty, harmful behaviour and violations of others' rights or trust.
05 / Accountability
Raising a concern
If you experience or witness conduct that may conflict with this Code, you can contact Sauti Labs to raise the concern. Share relevant details where you can, and avoid circulating sensitive information more widely than necessary.
We ask everyone to approach concerns honestly and respectfully. Retaliation against someone for raising a concern in good faith is not acceptable.
Contact Sauti LabsResponding to violations
Concerns should be considered thoughtfully, with attention to the available information, the people affected and the context of the situation. Responses should aim to address harm, protect participation and uphold the standards described here.
Appropriate action depends on the circumstances and the responsibilities and authority available to Sauti Labs. This Code does not replace applicable law or other relevant organisational policies.
A shared standard
Building useful, trustworthy African speech and language technology takes more than technical skill. It requires intellectual honesty, respect for people and communities, and a willingness to take responsibility for our work.
Related commitments