FundrAIsing: the Guidance on AI and fundraising

The rise of Artificial Intelligence (AI) has been exponential, and its effects have been far-reaching. For charities, the efficiency and ease of AI can assist with fundraising tasks such as streamlining fundraising data, generating custom posters or communications for events, or presenting models of donor behaviour to better understand patterns in donations. It comes as no surprise that following the prevalence of AI, the Fundraising Regulator released guidance on how to use AI in fundraising and remain compliant with the Fundraising Code of Practice (Guidance).

In this article, Jason Sinclair, trainee solicitor in our sport, education and charities team, provides a breakdown of the guidance and what it means for charities and their fundraising activities.

What are the risks of using AI in fundraising?

As a charitable organisation, it’s important to understand some of the legal and commercial risks associated with using AI in fundraising.

Legal risks

Data protection is a big part of fundraising. Charities will often collect the names, addresses and payment details of donors. If charities input this data into an AI model, which pushes data into and pulls data from the public domain as it learns, charities could be putting the personal data of their donors at risk. This is especially true of free-to-use versions of popular AI models like ChatGPT.

Charities should also be cautious about the possible infringement of existing Intellectual Property rights. Generative AI in particular could create a scenario in which a communication uses the reputation attached to pre-existing brands, images or words to make a fundraising event appear connected to a well-established charity.

AI is not legally trained, meaning any questions or outputs connected to legal rules must be checked for legal accuracy. Charities should refrain from asking AI models for fundraising legal advice. The best sources of information in relation to legal updates are those published by law firms, like this one, as their accuracy has been verified and backed up by legal experts.

Commercial risks and AI hallucinations

Hallucinations are a phenomenon whereby an AI doesn’t know the correct answer to the prompt but must answer it nonetheless so will make up, or ‘hallucinate’, a response to be able to satisfy its user. Public examples include ChatGPT providing a list of fictional cases to a barrister to use in court or an AirCanada AI chatbot inventing a refund policy to a customer.

Many AI models operate ‘behind closed doors’ meaning their logic or algorithm is not open to inspection. This can, and has, led to bias in AI models such as racism or sexism. When using AI to fundraise, be sure to sense-check its responses for any signs of discrimination.

How should trustees use AI in fundraising?

The Guidance highlights that trustees are responsible and accountable for a charity’s use of AI and they should be comfortable with the outputs of AI, considering the risks discussed above. There are certain approaches that the Guidance encourages charities to adopt to remain compliant with the fundraising code.

• Engagement - Trustees should be engaged with the strategic decision to use AI and monitor its continued use. It’s important that charities don't leave the responsibility to a few selected trustees, but rather engage as much of the board as possible. This increases awareness and knowledge of AI so trustees can make informed decisions. It also means that trustees are in a good position to identify when third parties they engage for fundraising efforts may be using AI or to spot ‘bad actors’ in the AI world.
  
  
• Risk Assessing - Good trustees will ensure that before any AI is used in fundraising, a risk assessment is carried out to understand its effects and possible consequences. While this risk assessment should be proportionate to the charity’s intended level of use of AI, it is important trustees are aware of the risks before it is utilised.
  
  
• Human Oversight - Trustees should put a process in place to ensure all outputs of AI are checked by a person to assess accuracy, fairness and, where necessary, the legality of its output. This minimises the chances of a hallucination or bias coming through. It may be hard to check every AI output, i.e. if there is an AI chatbot to manage fundraising correspondence, but there should be a human element to it, such as audits or ongoing monitoring. To do this,  there should be people adequately trained to implement, manage and oversee the AI tool, and if a charity has engaged trustees early in the process and conducted a risk assessment, it should be easier.

How to use AI responsibly in fundraising

Finally, the Guidance discusses using AI in fundraising and what charities should do.  There are four key takeaways listed below.

• AI Policy -  Charities should implement an AI policy before any tool is deployed either as a standalone fundraising AI policy or as part of a larger policy. The Guidance recommends this policy should be made publicly available for donors to understand when and how AI is used, for example in generating content or processing personal data. Data protection points should be seriously considered including understanding how the AI tool will interact with other IT systems and how data that is inputted, or outputted, is stored. Regardless of its use, the policy should be regularly reviewed.
  
  
• Testing AI - The Guidance recommends that any AI is tested before it is rolled out and the findings of the test are reported to the board of trustees. This would inform any risk assessment or highlight any risks potentially unknown at the time.   
  
  
• Training AI - Once tested, the AI tool should be trained. The Guidance recommends that charities do not feed information not in the public domain into publicly available, or open access, AI models; these are models like Chat GPT. The Guidance also recommends against using AI to make automated decisions related to fundraising. Nonetheless, specialist advice should be sought when it comes to legal accuracies.   
  
  
• Transparency on AI - Throughout the process, charities should be transparent with their donors when AI is being used in fundraising activities. This helps to develop public confidence and donor support for the safe and fair use of AI in fundraising.

The importance of consulting legal experts

Ultimately, the final line of the Guidance provides the overarching message, that decisions “based on AI content must be in the best interest of [the] charitable fundraising institution and uphold the reputation of fundraising.”

Whilst understanding the basics is key to ensuring better decision-making when it comes to using AI for fundraising purposes, seeking specialist legal support is always encouraged to mitigate risks to your charity’s position and reputation.

For support with using AI as a charity or non-profit organisation, contact Tom Justice via [email protected] or 0191 211 7913