AI Policy at We Move Europe

WeMove Europe recognises the growing role of Artificial Intelligence (AI) in digital work and campaigning. Used responsibly, AI tools can help our teams work more efficiently, strengthen our output and multiply our impact.

At the same time, AI technologies present risks relating to bias, transparency, environmental impact, and misuse of data. This policy sets out how WeMove Europe team members may use AI tools in ways that enhance our work while remaining consistent with our organisational values.

We operate in a rapidly evolving digital environment, often facing well-resourced and technologically sophisticated opponents. To remain effective, we aim to responsibly adopt tools that increase efficiency and impact, without compromising our values. Our approach to AI is guided by these values:

• Equality: We actively review AI outputs for bias, harmful stereotypes or discriminatory assumptions.
• Sustainability: We aim to use digital tools responsibly and avoid unnecessary use of resource-intensive tools.
• Democracy: We prioritise transparency in the use of AI and avoid applications that could mislead audiences or manipulate public discourse.

Purpose of this policy

This policy aims to:

• Enable teams to benefit from the efficiencies AI technologies provide
• Ensure the use of AI is aligned with WeMove Europe’s values of equality, sustainability and democracy
• Provide clear guidance on acceptable, restricted and prohibited uses of AI tools

Scope of the policy

This policy applies to all WeMove Europe staff and teams.

The policy should be reviewed at least once a year to reflect technological developments and organisational needs. It is owned by CMT and any questions or suspected misuse of tools should be reported to a member of CMT.

Data and confidentiality

All staff must continue to comply with WeMove Europe’s existing data protection and GDPR policies when using AI tools. They should also be logged into an approved private organisational account where available, instead of using anonymous or public sessions.

Staff must not input sensitive or confidential information into public AI systems unless the tool has been explicitly approved for secure internal use. Examples of information that should not be uploaded to public AI tools include:

• Personal data (staff, supporters, donors or partners)
• Email addresses or supporter databases
• Financial information (such as IBANs)
• Internal strategy documents
• Campaign plans
• Confidential organisational communications

Team members should avoid unnecessary or excessive use of resource-intensive AI tools, in line with our focus on sustainability. An example of unnecessary or excessive use might be generating large volumes of unused images/content.

Approved tools and uses

The following AI tools are currently approved for specific types of work:

Claude AI

• Analysing data
• Code and supporting coding tasks
• Testing and debugging code
• Translating copy

Chat GPT

• Translating copy
• Facilitating research
  • Research should not be delivered or instigated by Chat GPT. Instead it can be used as a powerful search engine tool for specific pieces of information. It can find and filter - it should not be used to steer or inform independently.
  • Example: Several references for the PFAS launch were taken down the day the copy was finalised. ChatGPT was used in this situation to find alternative reference sources, which were then reviewed and verified by two members of the campaigns team.

Canva AI

• Creating illustrative or stylised graphics (eg. collage-style images for our petition pages), which cannot be mistaken for real photographs or events

Any images generated using AI should be clearly labelled.

Team members who want to try a new tool or use it for a different purpose should seek approval from CMT before using it for organisational work.

The following uses are restricted/banned:

• Deceptive content - subjective but anything that could be viewed as something that did happen
• Uploading personal supporter data of any kind eg. email addresses
• Inputting confidential or sensitive company information into public AI tools, such as:
  • Personal data (staff, supporters, donors, partners)
  • Financial information
  • Internal strategy documents
  • Non-public campaign plans
• Relying on AI tools to analyse data or propose recommendations (while AI tools can be used to sort, filter and present data, team members are expected to be able to independently analyse data in their roles)

Broader/ethical commitments of our team

1: Always be aware of AI’s bias

A core value of WeMove Europe’s work is equality. It has been well documented that AI is not an objective source of information. Racial bias and other forms of discrimination are embedded into the technology due to the indiscriminate scraping of the internet that informs its intelligence. By drawing from the whole of the internet for its information, AI completely and uncritically reproduces the discrimination, privileging and bias that’s present there.

We’ll always have this front of mind when using AI, and will make sure we’re interrogating any results. While we know we won’t be able to eradicate bias, just as we can’t completely eradicate our own, we will work hard to ensure we don’t reproduce it without interrogation.

Staff should be alert to outputs that reinforce stereotypes or discriminatory language.

Outputs should be reviewed to ensure they do not disadvantage or misrepresent groups based on identity, background, or status.

2: Human oversight and accountability

While we think that AI has a lot of potential for opening up the creative process and innovating the digital campaigning offering, it cannot do this on its own. Our skilled team will always be producing copy, translating results, questioning outcomes, and building on any output of AI before sharing our work with partners and the public.

3: Transparency and disclosure

We do not anticipate producing published content where AI has significantly contributed. However, we commit to being transparent if such an instance occurs. We will actively avoid using AI tools to mislead audiences or obscure authorship. We uphold transparency as an essential pillar of a democratic society.