ELM Frequently Asked Questions (FAQs)

Here you will find answers to common questions and guidance on using our services. For further assistance, contact our support team.

At Edinburgh University, we recognize the environmental concerns associated with the use of Generative AI and are committed to reducing our digital carbon footprint. Our goal is to ensure that ELM provides a sustainable and responsible avenue for accessing Generative AI. To achieve this, we are implementing the following actions: 

 

Transition to Energy-Efficient Hosting: 

As of Q1 2025, ELM exclusively provides access to large language models from OpenAI. We are finalizing plans to integrate GAI models hosted on the University's energy-efficient data centre infrastructure. This transition will allow us to deploy smaller, less power-demanding models—thereby reducing our environmental footprint relative to leading external providers—and enable more accurate future measurement of ELM’s environmental impact. 

 

Enhanced Transparency and Measurement: 

While current limitations in data from third-party GAI providers prevent a full assessment of ELM’s environmental impact, our commitment to operating self-hosted models will facilitate a robust, baseline measurement of energy usage. This transparency will be essential in holding ourselves accountable and refining our sustainability practices over time. 

 

Responsible Use and Comprehensive Training: 

We encourage all users to remain mindful of the environmental implications of using ELM. To support sustainable practices, training is available from the Digital Skills, Design and Training Team that highlights which tasks are best suited to GAI and which could be completed using other systems, lowering the environmental cost. Details and sign-up information for these training sessions are available here.

 

Continuous Improvement and Guidance: 

Our sustainability journey is ongoing. We are dedicated to regularly expanding and updating our guidelines to promote efficient use of ELM and minimize the environmental impact of individual usage. These guidelines will evolve as new technologies and strategies emerge. 

 

Edinburgh University remains steadfast in its dedication to environmental stewardship. Through these strategic initiatives, we aim to lead by example in the responsible and sustainable deployment of Generative AI.


ELM operates under a ‘zero data retention agreement’ with OpenAI, the provider of the commercial Generative AI models that are used in ELM. This means your data is not retained by OpenAI and is secure. 

ELM will also soon offer access to non-proprietary ‘open source’ Generative AI models hosted locally and securely completely on the University’s data centres, which will mean all data associated with the use of the locally hosted Generative AI models will be secure within the University’s data centres. 

ELM, as with other IT systems in use at the University, operates under the Use of Operational Data Policy. You can access this Policy here. 


No, ELM has a Zero Data Retention policy in place with OpenAI. Your chats and uploaded data are not used by OpenAI for the purpose of training their models and are not retained or used by OpenAI.


By default, users can access their own chat history and prompts directly within their ELM instance. Future developments may offer additional functionality for exporting chat histories, but this will remain under the sole control and discretion of the user.


ELM does not make assertions or claims with regard to any user’s intellectual property.


- Right of Access: Chat histories and prompts are directly accessible via ELM. 

- Right to Rectification: Since ELM only holds discretionary user-supplied personal data (such as UUN), this Right does not apply. 

- Right to Erasure: This Right is qualified and will apply only if there is no legitimate reason to retain the data. 

- Right to Restrict Processing: ELM holds limited personal data beyond what users generate or upload themselves. This Right is self-exercised at the user's discretion. 

- Right to Data Portability: This applies to data processed under 'legitimate interest' or 'consent' and does not apply to ELM data. 

- Right to Object: Not applicable under ELM’s legal basis, which is the ‘performance of a contract’.


Access to ELM is (and was) enabled at the college or school level after consultation with key stakeholders within each department. Heads of College IT and/or Heads of School were responsible for approving access to ELM within their respective areas.


For more details on ELM, please refer to the ELM Guide . Additional information on AI more broadly at the University can be found on the Edinburgh’s AI page.

 

 


 

If you have any questions that remain unanswered, please feel free to get in touch!

edina@ed.ac.uk

 

This article was published on