AI image guidelines for charities 

More than three quarters of charities believe that authentic content helps organisations stand out and build trust, according to a new report. 

Yet despite this, 16.4% of charities are using images generated by artificial intelligence (AI) anyway, according to the ‘AI-generated imagery in the charity sector’ report, from charity filmmakers The Saltways.   

There are myriad risks to using AI-generated imagery as charities, including bias and misinformation. Indeed, The Saltways’ report was created in response to an investigation reported in The Guardian, which revealed that charities were experimenting with AI-generated images in their communications, potentially perpetuating harmful stereotypes around poverty.  

Saltways recognised the need for guidance around “how AI images are generated, what the legal requirements are, and what the potential risks are to their reputation and communities”. 

“Charities are filled with good people, with good intentions, doing good work and trying to do it ethically,” explains the report. “They didn’t need judgement, they needed guidance.” 

To fill the gap left by a lack of regulation and the continued evolution of AI technology, The Saltways have also created clear guidelines to help charities make informed decisions around whether to use AI-generated images to support their communications.  

It outlines the risks, sorting them into high and lower risk uses, and offers advice on mitigating those risks, including recommended platforms to use and disclosure best practice. It also features a checklist of questions to consider before using AI-generated imagery. You can find the full guidelines here.  

Below, we summarise some of the key findings in the report and what charities need to think about when it comes to responsible AI and imagery.  

AI-generated imagery in the charity sector 

Currently, there is no regulatory requirement in the UK for organisations to disclose if an image is generated by AI. However, organisations processing EU data are required to disclose this as part of The EU AI Act.  

The EU AI Act, which came into force in February 2025, states that “Deployers of an AI system that generates or manipulates image, audio, or video content constituting a deep fake, shall disclose that the content has been artificially generated or manipulated.” 

The Saltways’ report found that 94% charity professionals were unaware or unclear about these legal requirements. This statistic was particularly concerning given that many charities in the UK process EU citizen data and fines can be incurred to charities outside the EU if the risks impact those inside.  

Furthermore, even without the pressure of abiding by EU legislation, transparency around AI should still be important to charities, given the risks it poses to their trustworthiness. Yet 56% of respondents who are using AI images said they don’t consistently label AI-generated content, “either having no policy or labelling only rarely or never”. 

There is a connection here between the lack of transparency and the level of understanding when it comes to the risks of generative AI. The report found that charities are not very familiar with the ways that AI reinforces harmful stereotypes: “Whilst the majority had heard about these concerns, only about one in five had actually read research on the topic”. 

Many charities were also unaware of how AI works exactly. Yet without this knowledge, it is difficult for charities to properly judge the impact of using them. Just one in three charities said they had a strong technical understanding of how AI generates images, while more than half said they had a basic understanding – they knew it was computer generated but not the technical details.  

The Saltways believes that the charity sector has the opportunity to lead the way in ethical AI use. By acknowledging the risks of AI, and taking into the guidelines set out by The Saltways, charities can make more ethical decisions around technology and reward the trust put in them by their supporters and service users.  

You can find the full report and guidelines here.