Request for Proposal Independent Red Teaming and AI Safety Evaluation
Girl Effect
- Kenya
- Mid Level
- Posted
Skills
Required
- AI
- AI safety
- Kenya
- human escalation
- Kiswahili
- Sheng
- Africa
- Global South
- conversational AI
- retrieval-augmented generation
- multi-turn systems
- model evaluation
Preferred
- AI Safety Evaluation
- taxonomy
- API
- GenAI
- threat-modelling
- autonomy
- tool use
- modes
Procurement Summary
- Procurement Method Open competitive tender
- Assignment Type Independent organisation or specialist consultancy
- Contract Period Approximately 10 weeks
About Girl Effect
Independence is a mandatory condition. The selected evaluator must not be involved in building, operating, or commercially benefiting from the systems evaluated under this assignment.
Girl Effect is an international non-profit that builds media and technology that adolescent girls want, trust and need. We create digital-first brands and products that reach girls at scale across Africa and Asia, supporting them to make informed choices about their health, education and futures — with a particular focus on sexual and reproductive health (SRH).
Girl Effect has worked in responsible AI for several years and has published its own Ethical AI Guidelines and a Generative AI Whitepaper, co-authored sector guidance on the safe use of chatbots for social and behaviour change, and operates named accountability for AI safety — including a Failure Mode Register with safeguarding sign-off, and defined escalation protocols with service-level commitments for human follow-up.
We believe that AI serving young people in the Global South must be held to the highest evidential standard, and that the organisations building such systems should not be the only ones marking their own homework.
This engagement exists to put that principle into practice.
WAZZII is Girl Effect’s WhatsApp-based chatbot in Kenya. It provides adolescent girls and young women with confidential and non-judgemental information about their bodies, relationships and health. It also connects users to relevant products, referrals and health services, supported by a human help desk responsible for moderation and escalation.
WAZZII’s next phase will expand the use of generative AI in the girl-facing experience. This includes:
Responsibilities
- Item Details
- RFP Title Independent Red Teaming and AI Safety Evaluation
- Reporting To Head of Technology and Product
- Submission Email
[email protected]
- AI-assisted tools for human help-desk operators;
- Safeguarding and clinical escalation pathways;
Internal evaluation of model outputs; and
- Defined policies covering content and actions that the system must never produce.
- The objective is to conduct a focused and independent red-team evaluation of WAZZII’s girl-facing generative AI capability.
- Identify safety, safeguarding, clinical and technical vulnerabilities;
- Test whether the system recognises and safely escalates relevant risks;
- Assess whether existing evaluations and controls adequately reflect real-world use;
- Produce reproducible test cases that Girl Effect can rerun;
- Rank findings by severity, likelihood and potential impact; and
- Provide practical and prioritised recommendations for remediation and future evaluation.
- The core assignment is a single time-bound evaluation. Remediation re-testing, review of agentic designs and periodic testing will be priced separately as optional extensions.
- The core engagement runs over an indicative ten weeks.
- All work follows Girl Effect’s Ethical AI Guidelines, data policy and safeguarding policy: testing is conducted against staging environments using synthetic personas, no real user data is accessed or collected, data collection is minimised throughout, and no safeguarding case content is handled at any point.
- Girl Effect will provide staging access, including programmatic (API) access to the conversational endpoint, supporting automated test harnesses; testing may be conducted fully remotely, with no in-country presence required.
Key activities include
- Phase 1 — Orientation and test design (indicatively weeks 1–3)
- Focused desk review of Girl Effect’s evaluation framework, red-line policies and escalation protocol — sufficient to design a well-targeted test plan, not a standalone audit
- Co-scoping of the scenario taxonomy and severity rubric with Girl Effect’s AI Ethics Officer and safeguarding leads, and agreement of the responsible-disclosure protocol for critical findings
Phase 2 — Red-teaming sprint (indicatively weeks 3–8)
- A structured, tool-supported testing exercise against the girl-facing GenAI experience in staging, combining adversarial and realistic non-adversarial scenarios and producing logged, machine-readable, re-runnable test cases, prioritising: clinically incorrect or harmful health advice; failure to recognise and escalate safeguarding disclosures; jailbreaks, prompt injection and policy circumvention; and age-inappropriate content
- Testing across the languages and code-switching patterns of real users (including Kiswahili and Sheng); local-language competence may be provided by the partner or in pairing with Girl Effect's own team
- Stress-testing of the human-escalation pathway, primarily with realistic non-adversarial scenarios (genuine distress, disclosure and clinical-concern personas where the system should hand over without being attacked): does it hand over to the help desk when it should
- Phase 3 — Findings and briefing (indicatively weeks 8–10)
- A severity-ranked findings report with reproducible test cases, handed over so Girl Effect’s team can re-run them continuously after the engagement ends
- A findings briefing for Girl Effect leadership, with prioritised recommendations — including what a fuller evaluation should cover next and at what cadence
- Optional priced extensions (quoted separately, commissioned at Girl Effect’s discretion)
Re-test of remediated critical findings
- Architecture-level adversarial review of agentic system designs (threat-modelling autonomy, tool use and failure modes before deployment)
- Periodic re-testing and pre-launch adversarial gates for major releases
- Deliverables
- Deliverable
- Indicative timing
- Inception note & test plan
- Confirmed methodology, scenario taxonomy and severity rubric, staging/access arrangements, ethics and data-handling approach, responsible-disclosure protocol
- Red-team findings report
- Severity-ranked, reproducible findings from the adversarial sprint, including escalation-pathway results; test cases handed over for continuous internal re-use
- Leadership briefing & recommendations
- Findings briefing with prioritised recommendations and proposed scope/cadence for future independent evaluation
Duration and Working Arrangements
- The core engagement is expected to run for approximately 10 weeks.
- Programmatic access to the conversational endpoint where available;
- Relevant documentation and evaluation materials;
Named technical, safeguarding and programme focal points; and
- Consolidated feedback on deliverables.
- Participate in a weekly progress meeting;
- Maintain a shared delivery tracker and risk register;
- Use secure communication and evidence-sharing channels;
- Escalate critical issues immediately;
- Protect access credentials and evaluation materials; and
- Manage its personnel and delivery approach independently.
Independence and Conflicts of Interest
- Independence is a mandatory condition of award.
- Understanding of the assignment;
- Proposed methodology and testing tools;
- Proposed scenario taxonomy and coverage;
- Approach to automated and manual testing;
- Approach to safeguarding and escalation testing;
- Independence and conflict-of-interest declaration;
Proposals will be assessed in the following order
- Pre-screening;
- Technical evaluation;
- Financial evaluation.
- Evaluation Criterion
Procurement Timelines
- The consolidated FAQ will be shared with all prospective bidders through the same publication or communication channels used for the RFP.
- Bidders are responsible for reviewing the FAQ and any formal tender updates before submitting their proposals.
- Submit the technical and financial proposals as separate attachments to:
- [email protected]
- The agreed implementation pricing will remain fixed unless a change is approved through the formal written change-control process.
Requirements
- Clarification Deadline 10 November 2026
Proposal Deadline 18 November 2026
- Testing may be completed remotely. No in-country presence is required unless separately agreed.
- Approved access to the staging environment;
- Have no commercial interest in the evaluation achieving a particular outcome;
- Disclose any relevant relationship held during the previous 24 months;
- Experience independently red-teaming deployed AI systems rather than only evaluating standalone models.
- Experience with conversational AI, retrieval-augmented generation and multi-turn systems.
- Experience in health, safeguarding-sensitive or other high-risk contexts.
- Experience working with adolescent or youth-facing digital services.
- Strong knowledge of AI safety, responsible AI and model evaluation.
- Ability to build automated, machine-readable and rerunnable testing harnesses.
- Understanding of human escalation and moderation workflows.
- Ability to test English, Kiswahili and Sheng directly or through a credible quality-controlled approach.
- Experience working in Kenya, Africa or comparable Global South contexts.
- Strong security, confidentiality and responsible-disclosure practices.
- Excellent analytical, technical-reporting and leadership-presentation capability.
- The proposed team should include named individuals with appropriate AI evaluation, technical, safeguarding and language expertise.
- Proposal Submission Requirements
- Bidders must submit separate technical and financial proposals (Max 10 pages).
Technical Proposal
- Language and cultural-context capability;
- One redacted example of a relevant findings report.
- Financial Proposal
- The financial proposal must be submitted in USD and include:
Proposed team and safeguarding or AI-safety capability: 10%
- Only reasonable expenses approved in writing in advance will be reimbursable. Reimbursable expenses must be supported by appropriate documentation.
- Bidders must disclose any actual, potential or perceived conflict of interest.
Preferred
- The lowest-priced proposal will not necessarily be selected.
Pay
- A cost breakdown by phase and deliverable;
- Financial proposal and overall value for money 30%
Benefits
Primary Market Kenya
- A fixed price for the core 10-week engagement;
- Tools, licences and third-party costs;
- Reimbursable expenses shown separately;
- Retrieval-grounded responses using clinically approved content;
- Girl Effect seeks an independent partner to test whether these safeguards perform as intended and identify risks that may not have been detected through internal evaluation.
- The selected partner must be separate from the organisations developing or operating the systems being tested.
- Objective
Week 3
- Week 8
- Week 10
The bidder must
- Have no current contract to build or operate the system being evaluated;
- Disclose any current or previous relationship with Girl Effect’s AI, platform or technology suppliers;
- Identify any tools or models being evaluated in which it has a financial interest; and
- Explain how any potential conflict will be managed.
- Girl Effect will determine whether a disclosed relationship affects the bidder’s eligibility or independence.
- Data-handling and information-security approach;
- Responsible-disclosure process;
- Workplan and team structure;
- Relevant experience and comparable assignments;
- Key assumptions, dependencies, limitations and exclusions;
Three client references; and
- Personnel roles, daily rates and level of effort;
- Applicable taxes shown separately;
Day rates for each optional extension; and
- Clear assumptions, exclusions and conditions.
- Evaluation
- Pitch, demonstration or clarification stage, where required; and
Relevant AI red-teaming and system-evaluation experience: 25%
- Quality, coverage and practicality of the proposed methodology: 25%
- Language, local-context and affected-community perspective: 10%
- Technical Total 70%
Total 100%
Girl Effect may request written clarifications, invite shortlisted bidders to a pitch or technical demonstration, conduct reference checks and due diligence, or request a best-and-final offer.
Terms of reference published
4 November 2026
Questions/Clarification Deadline
10th November 2026
Consolidated FAQ shared with all prospective bidders
13th November 2026
Deadline for proposal submission
18th November 2026
Clarification Process
Bidders may submit clarification questions to [email protected] by 10th November 2026
The email subject should state
- Clarification Question –
- Independent AI Red-Team Evaluation
To ensure a fair and transparent open-tender process
- Girl Effect will not provide substantive answers to all bidders.
- All relevant questions and responses will be compiled into a consolidated FAQ.
- The FAQ will not identify the organisation that submitted each question.
- Verbal information or guidance will not amend the RFP unless confirmed by Girl Effect in writing.
- Girl Effect may decline to answer questions received after the clarification deadline.
Applications must be received by
18 November 2026
Use the email subject
Independent Red-Teaming and AI Safety Evaluation
- Late, incomplete or incorrectly submitted proposals may not be evaluated.
- Girl Effect may contact only shortlisted bidders. Unsuccessful bidders will be notified following completion of the procurement process.
- General and Commercial Terms
Contracting
The successful bidder will sign Girl Effect’s standard services agreement, statement of work and applicable data-processing and information-security documentation.
Payment
Payments will be linked to the satisfactory completion and written acceptance of agreed deliverables. Girl Effect may withhold payment for incomplete, defective or unapproved work.
Intellectual Property
Girl Effect will own all foreground intellectual property, deliverables, source code, configurations, designs and documentation created specifically under the engagement, subject to agreed treatment of pre-existing tools and third-party materials.
Confidentiality and Data Protection
The partner must protect Girl Effect’s confidential information and personal data and comply with applicable data-protection requirements.
Safeguarding and Conduct
Appointment is subject to satisfactory due diligence, reference, safeguarding, information-security and compliance checks. The successful bidder must comply with Girl Effect’s code of conduct, safeguarding and anti-fraud requirements.
Subcontracting
No material part of the assignment may be subcontracted without Girl Effect’s prior written approval. The lead bidder remains responsible for all subcontracted work.
Warranties and Defects
Bidders should propose a warranty period during which defects in accepted deliverables will be corrected at no additional cost.
Tender Costs
Girl Effect will not reimburse costs incurred in preparing or submitting a proposal or participating in clarification, pitch or negotiation stages.
Right to Award
Girl Effect may clarify, negotiate, amend, suspend or cancel the procurement; reject any or all proposals; award the full or partial scope; award in phases; or make no award.
No Commitment
Publication of this RFP does not constitute a commitment to enter into a contract.
Equal Opportunity
Girl Effect welcomes submissions from qualified organisations that meet the stated regulatory, technical and service requirements.
Proposal Validity
- Proposals must remain valid for 90 days from the submission deadline.
- Annexes
- Annex A —
- Girl Effect Ethical AI Guidelines
- Annex B —
- Girl Effect’s Generative AI Whitepaper
