How GiveDirectly prevents, addresses, and reports risks

• Delivering your donations as cash to people in poverty is life-changing, but it also creates risk of fraud and harm to recipients, their communities, and our staff.
• Below we outline how common those risks are for GiveDirectly and other aid organizations, plus our systems for preventing and responding when incidents occur.
• To improve our accountability and transparency, we’ll start publishing high-level annual incident reporting early next year.

We’re improving how we report to you on risks

Giving cash directly is a proven and dignified way to relieve multidimensional poverty. While your donations do a great deal of good for families in need, delivering them also creates a risk of harm and fraud, which we outline in this blog below. 

You should also be able to review GiveDirectly’s org-wide incidents and outcomes, but are currently unable to, as our public reporting has been largely ad hoc and program-specific. 

To help others learn from our work and raise our bar for transparency, here’s how we’re improving:

• Starting this year, we’ll release an annual summary of org-wide incidents of fraud, harm, and safety/security—look for it linked here in Q1 2025. 
• In subsequent years, we plan to continuously improve the detail and specificity of this reporting, including data analysis on which programs and contexts are reporting the most risk.

Providing aid can result in physical, sexual, and psychological harm

We have a zero-tolerance policy towards anyone who harms another person. We’re continuously improving our systems and culture to prevent harm from happening, detect and respond quickly when it does. 

Despite these efforts, in our 15 years of delivering cash to 1.6 million people, we’ve dealt with a range of cases, including:

• workplace harassment
• sexual exploitation and abuse
• physical violence
• financial exploitation
• outside security threats (Appendix)

Such incidents happen in all large aid organizations, and each type of program creates unique risks. For our work, these include: 

• Distributing life-changing cash to people living in extreme poverty creates a power imbalance between our staff/partners and those we serve that is opportune for harm.
• Similarly, many of our staff live in countries with few employment options, creating a greater power imbalance between them and their managers.
• Giving sizable amounts of money to families or communities can, in rare instances, create or worsen existing interpersonal conflict, though research shows it more often reduces it.

Aid groups underinvest in prevention and fail to report incidents

The frequency of harm in the wider aid sector is largely unknown and approaches for addressing it are lacking. Below we quote from the UK’s 2018 Sexual exploitation and abuse in the aid sector report, but these points apply to all forms of harm:

• “Due to confirmed under-reporting, the exact scale is currently impossible to define, but practitioners suspect that those cases which have come to light are only the ‘tip of the iceberg’… 
• “Improving reporting of sexual exploitation and abuse is vital to understanding the problem, responding to it, and ultimately, to preventing it… 
• “Reports of sexual exploitation and abuse by aid workers and/or peacekeepers have emerged, the sector has reacted, but then the focus has faded… We heard from aid organisations that this was due to the pressure to reduce overheads.”

GiveDirectly’s safeguarding team works to prevent and respond to harms

In the past two years, GiveDirectly has increased our investment in safeguarding. Today, we have a dedicated safeguarding manager in each country we work in and global leadership from our VP of Risk, Internal Audit, Safeguarding and Compliance. We also contract with outside experts for specific system improvements or investigations. 

With our compliance team, safeguarding staff prevent harm by:

• Improving program design to minimize opportunities for harm (see Appendix); 
• Closely vetting data sharing agreements;
• Providing mandatory, recurring staff training on policies, code of conduct and values;
• Setting background and reference checks standards for new hires;
• Sensitizing communities of the risk of harm and ways to report it.

Today, when we receive an allegation, we initiate an investigation. If this investigation confirms harm was done, we take appropriate action. Depending on the case, this might include disciplinary action for staff, redress for survivors, and/or referral to local law enforcement for perpetrators. We also share information on staff misconduct with other aid organizations, as appropriate, through a global misconduct disclosure scheme to prevent perpetrators from finding new roles to commit harm.

Our survivor-centered approach means we don’t share all details of sexual harm cases

While we’re committed to sharing org-wide incidents of harm in our 2024 annual reporting, there are certain case details we are not able to disclose as we prioritize a survivor-centered approach, putting first the safety and well-being of survivors of sexual exploitation, abuse, or harassment (SEAH). 

“The survivor’s wishes, safety, and well-being remain a priority in all matters and procedures… Investigations must uphold confidentiality at all times… Information relating to the case should be shared only on a limited, need-to-know basis… Decisions should be made in the best interests of the survivor, rather than the organization.” – Guidance on a survivor-centered approach from the Core Humanitarian Standard Alliance

Most survivors ask us to keep their experiences and testimony confidential, only sharing with staff immediately working on the investigation. We do not publish specifics that go against their wishes or could negatively impact their safety or well-being. More so than fraud or other types of harm, publication of sexual cases can bring stigma to survivors, which has been known to result in more harm. 

Ultimately, our goal is to prioritize a survivor-centered approach, while still acting on our values of transparency and candor.

Fraud can occur via diverted transfers, business fraud, or outside scams

Fraud loss is common for all nonprofits but generally accounts for a fraction of their total budget: the former head of counter-fraud for Oxfam UK says, “Estimates of the scale of loss to fraud are between 2% and 5% of an organization’s income.”

GiveDirectly delivers over $150M a year, and, to date, our systems find our org-wide annual losses due to fraud have been consistently at or under 1%, but could increase as our detection improves. 

GiveDirectly funds can be lost to fraud when staff or outsiders:

• Divert recipients’ cash transfers before they receive it (SIM swapping).
• Enroll ineligible people to receive a GiveDirectly cash transfer (imposters). 
• Take recipients’ cash transfers or property bought with those transfers (bribes or theft).
• Commit business fraud.

Additionally, people in need can be targeted by scammers pretending to be GiveDirectly who ask for money in exchange for enrollment, which we work to prevent and address (e.g., WhatsApp scams). 

Digital cash aid is more resistant to diversion fraud than other aid

Consider a program giving food aid to people in poverty: the procurement process of finding a food producer is open to corruption, and the long journey the food takes from manufacturer to recipient via warehouses, distribution centers, and sub-grantees is open to diversion at every step.

In a GiveDirectly program, we typically run our process end-to-end as money is digitally transferred from our bank account directly to a recipient’s mobile money SIM card without intermediaries. Both the recipient and GiveDirectly know and can verify the exact amount transferred. While there are comparatively fewer variables to monitor in our programs, we still carefully protect against fraud.

We stop fraud with a unique firewalled field audit team and machine-learning data flags

All of our teams work to mitigate losses from fraud, but GiveDirectly is unique for having a firewalled internal audit team dedicated to ensuring safe cash delivery. Since 2016, this fully independent team has run randomized 1:1 follow-ups to check the work of our enrollment teams, surface any inconsistencies or concerns from community members, and conduct investigations. When needed, we involve local authorities in investigations and redress, working to recover stolen funds. 

This type of team is exceedingly rare in the nonprofit sector. From discussions at roundtables with other aid organizations, we’ve found many have financial audit teams looking for incongruities in spreadsheets and receipts, but almost none have on-the-ground, anonymous teams that audit their field operations.¹

Beyond our internal audit team’s follow-up visits, we’re continuously enhancing our other prevention systems to quickly spot and respond to fraud attempts. Starting last year, we:

• Increased interactions between our senior staff and local community members/leaders and upgraded our whistleblower reporting platforms.
• Conduct more advanced detection analysis, automatically flagging anomalies in field operation data for investigation through machine learning.

Stopping harm and fraud comes with tough tradeoffs

While it would be heartening to hear that, “GiveDirectly is doing everything possible to prevent risks,” that would not be fully true, as this work brings tough tradeoffs with no obvious right answer: 

• Speed: After a deadly earthquake, do you launch a relief program in 1 month to allow for hiring a robustly background-checked and trained team as survivors suffer or in 1 week with fewer preventive staffing processes but potentially save more lives? 
• Targets: How do you set safeguarding and audit teams’ targets? “Prevent all cases” deters reporting. “Resolve X cases a quarter” incentivizes rushing through investigations.
• Transparency: Publicly sharing details of harm cases helps the aid sector improve but could put survivors at risk and go against their wishes. How much do you share and when? 
• Acceptable risk: How do you define an avoidable harm vs. an unavoidable one when working in inherently high-risk areas like conflict zones?
• Costs: Fraud loss prevention and recovery can cost more than what is saved. Should our goal be 0.00% loss? If not, how do we calculate the “sweet spot” for cash leakage?

Our first value of “recipients first” guides our decision making on these trade-offs, but how we navigate them will change over time. Addressing risk is a moving target, ever-changing as our systems improve, our work grows to new contexts, and bad actors change their tactics. We aim to update you on our lessons learned and evolving approach in our annual reporting.

We aim to raise the bar on how risk is addressed and reported

We’re raising our standards in communicating on these risks and innovating in how to prevent and address them. This is vital to better serve GiveDirectly recipients and staff, our donors, and the wider aid sector.

Some aid groups under-invest in addressing harm and fraud and under-report it to the public, focusing instead on saving reputations and costs. Yet, there are also useful best practices from the sector (UN, CHS, FCDO) and commendable examples on reporting (Save, PLAN, UN).

The more nonprofits report on their incidents, failures, and innovations, the better and more accountable all aid will become. GiveDirectly will be part of that change.

Main text footnotes:

1. Notable exceptions include the I.R.C. and the U.N., which run robust ethics and compliance audits and share their reports publicly. ↩︎