Skip to main content

B10: Use of AI in Elections

The following table describes the status, the proposer and other metadata of this motion.
Metadata of the motion
Consultation:Winter General Meeting 2026
Agenda item:3. Motions of Policy and Organisation
Proposer:Oliver Ashton (Green Party)
Status:Published
Submitted:01/23/2026, 23:59

Motion text

1.Candidates are discouraged from using generative artificial intelligence

(AI) within their campaigns. This includes campaigning material, social

media posts, speeches, official communications and their candidate

statements.

2. Any material, media or communications should be made by candidates or

their supporting campaigners. Candidates who are believed to have used

generative AI in their campaigns will be asked by the ERO to remove and/or

change the material, media or communications. Any member can raise

concerns over use of AI in campaigns by candidates and it is up to the ERO

to investigate the claims.

3. The ERO may identify possible AI usage through: irregular wordings and

anomalous language, images with known traits of generative art or by other

means the ERO deem fit. The ERO may request proof that the media was

created by the candidate.

4. A record of rulings made by the ERO should be kept by the DAC with the

alleged candidate named in the record and the whistleblower left

anonymous. These records shall be made publicly available to all members.

Reason

To discourage the use of AI in internal elections. The membership deserve to know that the campaign ran by a candidate is real and that policies by a candidate are of their own mind.

Go to line:

Supporters


  • Jemima Gayfer-Thoms
  • Alfie Neumann

Amendments

none

Comments

01/27/2026

Cynthia Muthoni:

Q: does this include subtitles and audio enhancement?
01/29/2026

Darius Seago:

Q: I've used AI to help clean up statement will this be classsed as friendly amendment if added.

I completely agree with us being against using AI to completely write a statement or a campaign message
01/29/2026

Manu Teague-Sharpe:

I support this motion in general. I'm a bit concerned about enforcement and issues that may arise in the future. As generative AI becomes more widely used and developed, it will be unclear which tools are generative AI. As an example, in Canva you can remove the background to any photo instantly. This is clearly some sort of AI, but probably not generative AI. But in the future it may be done by generative AI, and it may not be clear to the person that used it or the ERO. (Another example is subtitles to videos, as Cyn mentions)

Saying all this, I do support this motion and we perhaps change aspects of it in the future.
01/30/2026

Samuel Hall:

Hi, I’d like to comment on this motion from the perspective of someone who works professionally in AI and machine learning.

I strongly support the aim of fairness and authenticity in internal elections. However, I have serious concerns about how this motion could be enforced in practice.

The boundary between “AI-generated” and “human-created” content is becoming increasingly blurred. Almost all modern software now incorporates some form of AI, including widely used tools such as Google’s spellcheck, grammar correction, and accessibility features. This makes it extremely difficult to define what should and should not be considered “AI use” in a meaningful way.

Related to this, “spotting AI-generated content” is becoming increasingly difficult, and there is currently no reliable technical method for doing so. In practice, most detection relies on informal “rules of thumb” and personal judgement. These are highly subjective and vulnerable to bias.

Allowing candidates to be investigated, sanctioned, or potentially disqualified based on opinions such as “this looks like it might have been made using AI” sets a very dangerous precedent. It risks unfair accusations, reputational harm, and inconsistent enforcement.

Even experts in this field cannot reliably determine the origin of most digital content. Building a regulatory system around such uncertain evidence is therefore both impractical and unjust.

Rather than relying on unreliable detection methods, I would encourage a focus on transparency, honesty, and equal access to campaign resources, regardless of whether AI tools are used.