I want to be open and upfront with you about how we use artificial intelligence (AI) in my office.

We’re committed to using AI in a responsible, ethical and secure way. 

We know that when AI is used in the right way, it can often help get our work done quicker and more efficiently, improving the way we get things done for our community.

AI isn’t perfect, it makes mistakes too. That’s why we also understand how important it is that people - not just tech - maintain a watchful eye over any work that is produced by AI and used in our office.

Being open about this with you is an important way to build and hold on to public trust in our work.

This Artificial Intelligence Transparency Statement sets out:

• How we use AI tools within our office;
• The principles that guide that use; and 
• The things we do to make sure our practices are lawful, ethical and secure.

This statement will be reviewed from time to time to make sure it remains accurate, relevant and in line with best practices, legal changes and technological developments.

Our approach to AI

Our approach to AI is guided by the following principles:

• People, not tech, always make the final decisions on what is produced and released. It’s important to state, AI is used only as a support tool. Human judgement, decision-making and accountability matter at all times.
• We’re upfront with you: We are open about how and why AI tools are used within our office.
• Privacy and security matters: We prioritise the protection of personal, confidential and sensitive information.
• Lawful and ethical: AI will be used in line with legislation and parliamentary obligations.
• Only where needed: AI is used only where it adds value and does not replace appropriate professional expertise.
• Environmental considerations: We recognise AI technologies can have environmental impacts, including energy and resource consumption. We seek to use AI tools only where we have to or where it can help get work done faster. We do so in a measured and efficient way, avoiding unnecessary or heavy use.

We view AI outputs as guides only, not the final product of work. Decisions about what stands as final work products, advice or choices rests with our team, not AI systems.

How we use AI

Within our office, AI tools may be used in limited and controlled ways, including:

• Drafting assistance
  Help prepare early drafts or small sections of some simple, written material such as briefing notes, letters, longer emails or background papers.
• Research support
  Assisting with high-level research tasks, such as summarising publicly available information. 
• Proofreading and editing
  Helping to improve the quality of the work we prepare - for example, through better grammar and by making the work clearer, more readable, and easier to understand.

In all cases, AI-generated content is reviewed, checked and agreed by our team before use. AI tools are no substitute for local knowledge, or the care we have for what is in the best interest of our community.

We are committed to protecting privacy and confidentiality. When using AI tools:

• Only non-sensitive, non-identifiable information is entered.
• Everyone on our team will be responsible for making sure we follow the relevant privacy obligations, careful with how we handle and use information.

How we do not use AI

To safeguard privacy, security and public trust, our office does not and will not use AI:

• To input, process or analyse personal, confidential, classified or sensitive information, including constituent data. This includes names, contact details, and case-sensitive information such as health, financial or immigration-related information.
• To make autonomous decisions that affect individuals, policy positions or parliamentary work.
• To provide advice without appropriate human oversight.
• To replace human judgement in matters requiring careful thought, openness or ethical judgement.

How our AI use is governed 

Our AI use is governed through existing workplace arrangements, including:

• Follow parliamentary policies about ICT, information security, privacy and acceptable use.
• Limited device and network access, including the use of approved systems and platforms within the Parliamentary Computing Network. 
• Staff awareness and responsibility, ensuring those who use AI tools understand their limitations, risks and obligations.
• Ongoing review, as technology, policy settings and expectations grow and change.

Risk Management

We recognise that AI systems may produce inaccurate, biased or incomplete work. To lower the risks of this affecting our work, we make sure:

• All AI-assisted work is subject to human review and checking.
• We cross-check AI outputs against reliable sources, wherever we can.
• We monitor how technologies and the companies that own them are developing, to make sure we are up to date on any new risks associated with specific AI technologies.
• AI tools are not used in high-risk or sensitive contexts.