Constituent AI Policy

Constituents approaching my office with AI-generated correspondence may be asked to respond explaining the issue or what they hope to achieve in their own words.

We appreciate that this has caused some frustration, as many feel that they should be allowed to use AI to generate correspondence that feels more professional or polished, however, this policy has been created in response to many issues caused by this practice, including:

• Unfortunately, we are receiving a sizeable amount of fake casework aimed at overwhelming the office or misrepresenting public opinion. This can mean less time for us to deal with real, urgent casework or policy matters.
• AI generation often adds details which are inaccurate or unintended by the drafter in order to “pad out” the email.  Focus on these details when we make representations to external organisations on your behalf can cause confusion and put getting you a positive resolution at risk.
• We are frequently receiving AI-generated emails which are abusive – the default tone for some AI agents in writing to politicians seems to be aggressive.  While we understand the person behind it may not have intended or explicitly instructed this, please keep in mind that, on this end, real people – often staff – are reading your email.
• Part of our role in dealing with casework is problem-solving. AI often creates a list of demands which do not relate to real-world processes or people, or are otherwise unachievable or impractical.  Focus on these can prolong a case and cause frustration on both ends.

We do not mind if your emails are perfectly drafted or grammatical, or if you do not know what the exact solution looks like.

We are happy to correspond in different formats if required.  Please let us know of any access requirements you may have.

Staff AI Policy

My staff are also subject to an internal AI policy, detailing when it is appropriate to use AI tools in their work – and when it is not.  This covers all aspects of parliamentary working, including communications and policy briefings, as well as how to deal with constituents.

Specifically, the whole team is aware that it is not appropriate to hand constituent details to external AI agents to support in dealing with a case, and have been instructed not to outsource their thinking or empathy.

If you feel that AI has been used inappropriately in handling your case, please let me know and this will be reviewed.