Any use of AI technologies must protect, promote and fulfil human rights to the fullest extent. The technology must be assessed for its accuracy and transparency, and any decision to pilot, procure or use AI technologies must fully comply with the legal obligation in s. 38 of the Charter. AI Technologies must:
- consider how and where the systems are used.
- identify what if any data is being collected by the system and does it comply.
- seek community engagement.
- seek legal advice to ensure compliance.
- comply with the protected human right of privacy and reputation in s. 13 of the Charter.
The use of AI must have a demonstratable community benefit that clearly achieves one or more core objectives of the Victoria Police Corporate Strategy. Alternatives will be considered and, on balance, the benefits to community safety must outweigh any potential negative impacts.
AI products and systems:
- should deliver key insights into decision-making.
- must be the most appropriate solution for a service delivery or policing problem. The product or system should always be considered against other analysis and policing tools.
- should be preferred when they present maximum benefit to the community in alignment with Victoria Police Values and Code of Conduct.
- should deliver a clear community or government benefit and/or insight.
- must ensure the use of the product or system solution aligns with the Victoria Police strategic direction.
AI systems must be designed and used in a way that protects and promotes ‘human rights’, is transparent and fair and is non-discriminatory in its impacts on individuals and communities across Victoria.
AI systems must not be used to unjustly harm, exclude, disempower or discriminate against individuals, groups or communities. Potential harms and biases should be identified and managed, with guiding principles including:
- Use of AI products and systems should include safeguards to manage data bias or data quality risks.
- The best use of AI products and systems will depend on data quality and relevance. It will also rely on careful data management to ensure potential data biases are identified and appropriately managed.
- AI product and system solutions that rely on sub-optimal quality data may result in sub-optimal project outcomes and recommendations. Algorithms that contain systemic and repeatable errors may lead to prejudiced decisions or outcomes.
- Measures should be taken to ensure the AI produced decisions are compliant with the Equal Opportunity Act and the Charter.
- A use of data model that is appropriately tailored to the use case and as appropriate is reflective of community expectations on diversity and inclusion.
- Output to be reviewed to ensure that AI system achieves the required outcome.
Victoria Police is committed to protecting against the misuse, loss or unauthorised access and disclosure of personal information in accordance with the Information Privacy Principles under the Privacy and Data Protection Act (2014).
Privacy and security will be at the forefront of the design and use of AI systems. Victoria Police will comply with relevant privacy, data collection, use and disclosure, quality, security and information management and obligations.
- AI products and systems should strive for the highest levels of assurance.
- The Victorian community must have confidence that data used for AI products and systems is used safely and securely, and in a way that is consistent with privacy, data sharing and information access requirements, which includes the Privacy and Data Protection Act (2014), and Victoria Policy Manual (VPM) Information Privacy.
- Any AI product and system project outcome will be undermined by lack of public trust if there is any risk of a data breach, or that personal data could be compromised.
- Cyber security risks and vulnerabilities of AI systems and products are to be assessed and addressed.
- Where possible, an agreement on the consent for data use, with sufficient information provided on how the data will be used to ensure informed consent.
Victoria Police is committed to engaging, connecting and building trust with the community. We will provide transparency through collaboration and engagement.
The community must have assurance that data is being used safely and in accordance with relevant legislation and policy. They must also have access to information on Victoria Police use of AI products and system.
The community should wherever possible be informed of outcomes with an opportunity to ask questions about the use of data.
- Existing legislated oversight bodies will assist in providing community assurance allowing opportunity to question and challenge AI product and system outcomes.
- The development of AI solutions must be technically, legally, and ethically robust.
- Whenever possible, the community should be engaged on the objectives of AI products and systems.
- Victoria Police should ensure clear and understandable information about the use of the product or system is made publicly available to the greatest extent possible without undermining policing objectives.
- Victoria Police needs an understanding of how algorithms work and how decision making is informed.
- Should use community focus groups to ensure outcomes of AI systems fit within community expectations.
Victoria Police will ensure appropriate layers of governance at all stages to ensure it retains primary accountability for the AI system and the decision making it informs.
Victoria Police will continuously monitor, test and develop AI systems and ensure they are derived from relevant and contemporary data.
- Victoria Police will ensure it remains accountable for the use of the AI products and systems contracted through external vendors.
- AI products and systems are a powerful tool for analysing and looking for patterns in large quantities of data, undertaking high-volume routine process work, or making recommendations based on complex information. Any member using, administrating, or managing data in an AI system must receive an appropriate level of training (as defined in an associated VPM).
- No police powers or responsibilities should ever be exercised by an AI technology alone. There must always be a human decision-maker in any process involving AI technology.
- Victoria Police remains responsible for all AI product and system informed decisions and will monitor them accordingly. Testing, review and assurance processes of AI applications and systems will ensure optimal functionality and that AI systems continue to meet their intended objective.
- Assurance is required that AI products and systems are overseen and regularly audited by individuals with the relevant expertise, who monitor and report on risk.
Victoria Police will ensure that AI products and systems are only used to inform and enhance decision making, rather than to independently make decisions or determine outcomes.
- There should be appropriate human oversight and control at all stages of the development, deployment and operation of the technology.
- Human oversight will allow for decision making on the basis of human discretion.
- Victoria Police must always ensure that any AI technology it uses provides for the primacy of human decision making and discretion, to ensure fairness and accountability.
- AI products and systems are overseen by individuals with the relevant expertise and training in the technology its benefits and risks. Only those with required credentials will have access to data and systems.
- Datasets are to be reviewed by individuals with the training and ability to identify errors and remediate issues to improve accuracy.
Victoria Police will ensure members have appropriate training, skills and knowledge to develop deploy and operate AI products and systems.
This includes understanding the capabilities, risks and limitations associated with the system.
- A practice guide should be developed in conjunction with the implementation associated technology and policy.
- The Victorian community must have confidence that all employees using biometric technology are appropriately trained and have the correct authoritative powers.
- AI products and systems should only be used by trained persons who follow the Victoria Police endorsed policies and procedures. The level of skill or knowledge required will be dependent on role and the responsibility.
Updated