Automated decision-making and profiling Detailed Guidance

You are here:
Estimated reading time: 1 min
Information Commissioner’s Office, “Guide to the GDPR”, retrieved on 31st May 2018, licensed under the Open Government Licence.

These pages sit alongside our Guide to the GDPR and supplement the Article 29 Guidelines on Automated individual decision-making and profiling. They provide more practical guidance for UK organisations on automated decision-making and profiling under the GDPR.

The GDPR includes provisions to reflect organisations’ increasing use of profiling and automated decision-making across a wide range of applications. These provisions are designed to protect individuals from the potential risks that this type of processing can create.

The requirements are stricter when profiling is part of a solely automated decision-making process with legal or similarly significant effects.

This guidance will help you understand your responsibilities, what steps you need to take and why. It sets out how the ICO interprets the GDPR and our general recommended approach to compliance and good practice. It should be read alongside the more detailed Article 29 Guidelines detailed below.

For an introduction to the key themes and provisions of the GDPR, you should refer back to the Guide to the GDPR. You can navigate back to the Guide at any time using the link at the top of this page. Links to other relevant guidance and sources of further information are also provided throughout.

  • What’s new under the GDPR?

    In detail

    Yes, the GDPR introduces a new definition of profiling. It also restricts solely automated decision-making with legal or similarly significant effects.

    “The data subject shall have the right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning him or her or similarly affects him or her”

    Article 22(1)

    You can only carry out this type of automated decision-making when it is:

    • necessary for the entry into or performance of a contract;
    • authorised by Union or Member State law; or
    • based on the individual’s explicit consent.

    You also have a responsibility to:

    • actively inform the data subject about profiling and automated decision-making;
    • put suitable safeguards in place when you carry out this type of processing; and
    • introduce procedures for individuals to exercise their rights.

    Additional restrictions apply for special category data and automated decision-making involving children.

    Individuals had a right to be informed about automated decisions that significantly affected them but generally you could carry out this type of processing unless you received an objection. Individuals could use their rights under section 12 of the 1998 Act to:

    • serve notice requiring you not to take any automated decisions using their personal data; and
    • ask you to reconsider a decision taken by automated means.

    These rights didn’t apply if the effect of the decision was to grant the individual’s request.

    Large numbers of organisations now carry out profiling or take significant decisions about individuals by wholly automated means. These technologies can be used in ways that have a considerable impact on people’s lives.

    Organisations don’t just collect data directly from individuals, but from a wide range of other sources. This makes it less obvious to someone that their personal data is being used:

    • to profile them;
    • in a solely automated decision-making process; or
    • in ways that have an unanticipated effect.

    The GDPR provisions don’t just focus on reviews of automated decisions, or human involvement. They aim to address the risks from this type of processing by introducing new transparency requirements and higher protection for individuals.

    • Documenting your processing activities will help you identify whether your processing falls under the definition of profiling and automated decision-making.
    • If your processing involves profiling and significant automated decision-making you should carry out a Data Protection Impact Assessment (DPIA). This can help you decide whether Article 22 applies, the risks associated with the processing and how you are going to address them.
    • You must include specific details about this type of processing in your privacy information.
    • Ensure you have procedures in place to allow challenges from individuals and an independent review mechanism. You might be able to adapt any processes you already have in place to handle section 12 objections under the 1998 Act.
  • What is automated individual decision-making and profiling?

    In detail

    Profiling analyses aspects of an individual’s personality, behaviour, interests and habits to make predictions or decisions about them.

    The GDPR defines profiling as follows:

    ‘profiling’ means any form of automated processing of personal data consisting of the use of personal data to evaluate certain personal aspects relating to a natural person, in particular to analyse or predict aspects concerning that natural person's performance at work, economic situation, health, personal preferences, interests, reliability, behaviour, location or movements.

    Article 4 (4)

    Organisations obtain personal information about individuals from a variety of different sources. Internet searches, buying habits, lifestyle and behaviour data gathered from mobile phones, social networks, video surveillance systems and the Internet of Things are examples of the types of data organisations might collect.

    They analyse this information to classify people into different groups or sectors. This analysis identifies correlations between different behaviours and characteristics to create profiles for individuals. This profile will be new personal data about that individual.

    Organisations use profiling to:

    • find something out about individuals’ preferences;
    • predict their behaviour; and/or
    • make decisions about them.

    Profiling can use algorithms. An algorithm is a sequence of instructions or set of rules designed to complete a task or solve a problem. Profiling uses algorithms to find correlations between separate datasets. These algorithms can then be used to make a wide range of decisions, for example to predict behaviour or to control access to a service. Artificial intelligence (AI) systems and machine learning are increasingly used to create and apply algorithms. There is more information about algorithms, AI and machine-learning in ICO's paper on big data, artificial intelligence, machine learning and data protection.

    You are carrying out profiling if you:

    • collect and analyse personal data on a large scale, using algorithms, AI or machine-learning;
    • identify associations to build links between different behaviours and attributes;
    • create profiles that you apply to individuals; or
    • predict individuals’ behaviour based on their assigned profiles.

    Although many people think of marketing as being the most common reason for profiling, this is not the only application.

    Example

    Profiling is used in some medical treatments, by applying machine learning to predict patients’ health or the likelihood of a treatment being successful for a particular patient based on certain group characteristics.

    Less obvious forms of profiling involve drawing inferences from apparently unrelated aspects of individuals’ behaviour.

    Example

    Using social media posts to analyse the personalities of car drivers by using an algorithm to analyse words and phrases which suggest ‘safe’ and ‘unsafe’ driving in order to assign a risk level to an individual and set their insurance premium accordingly.

    Automated decision-making is the process of making a decision by automated means without any human involvement. These decisions can be based on factual data, as well as on digitally created profiles or inferred data. Examples of this include:

    • an online decision to award a loan; and
    • an aptitude test used for recruitment which uses pre-programmed algorithms and criteria.

    Automated decision-making often involves profiling, but it does not have to.

    Example

    An examination board uses an automated system to mark multiple choice exam answer sheets. The system is pre-programmed with the number of correct answers required to achieve pass and distinction marks. The scores are automatically attributed to the candidates based on the number of correct answers and the results are available online.

    This is an automated decision-making process that doesn’t involve profiling.

    Profiling and automated decision making can be very useful for organisations and also benefit individuals in many sectors, including healthcare, education, financial services and marketing. They can lead to quicker and more consistent decisions, particularly in cases where a very large volume of data needs to be analysed and decisions made very quickly.

    Although these techniques can be useful, there are potential risks:

    • Profiling is often invisible to individuals.
    • People might not expect their personal information to be used in this way.
    • People might not understand how the process works or how it can affect them.
    • The decisions taken may lead to significant adverse effects for some people.

    Just because analysis of the data finds a correlation doesn’t mean that this is significant. As the process can only make an assumption about someone’s behaviour or characteristics, there will always be a margin of error and a balancing exercise is needed to weigh up the risks of using the results. The GDPR provisions are designed to address these risks.

    Further reading

    The Article 29 Working Party includes representatives from the data protection authorities of each EU member state. It adopts guidelines for complying with the requirements of the GDPR

    WP29 Guidelines on automated individual decision-making and profiling – Chapter II

  • What does the GDPR say about automated decision-making and profiling?

    In detail

    The GDPR gives people the right not to be subject to solely automated decisions, including profiling, which have a legal or similarly significant effect on them.  These provisions restrict when you can carry out this type of processing and give individuals specific rights in those cases.

    Article 22(1) of the GDPR limits the circumstances in which you can make solely automated decisions, including those based on profiling, that have a legal or similarly significant effect on individuals.

    “The data subject shall have the right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning him or her or similarly affects him or her”

    Article 22(1)

    Solely means a decision-making process that is totally automated and excludes any human influence on the outcome. A process might still be considered solely automated if a human inputs the data to be processed, and then the decision-making is carried out by an automated system.

    A process won’t be considered solely automated if someone weighs up and interprets the result of an automated decision before applying it to the individual.

    Example

    A factory worker’s pay is linked to their productivity, which is monitored automatically. The decision about how much pay the worker receives for each shift they work is made automatically by referring to the data collected about their productivity.

    This is an example of solely automated decision making.

    Many decisions that are commonly regarded as automated actually involve human intervention. However, the human involvement has to be active and not just a token gesture. The question is whether a human reviews the decision before it is applied and has discretion to alter it, or whether they are simply applying the decision taken by the automated system.

    Example

    An employee is issued with a warning about late attendance at work. The warning was issued because the employer’s automated clocking-in system flagged the fact that the employee had been late on a defined number of occasions. However, although the warning was issued on the basis of the data collected by the automated system, the decision to issue it was taken by the employer’s HR manager following a review of that data.

    In this example the decision was not taken solely by automated means.

    A decision producing a legal effect is something that affects a person’s legal status or their legal rights. For example when a person, in view of their profile, is entitled to a particular social benefit conferred by law, such as housing benefit.

    A decision that has a similarly significant effect is something that has an equivalent impact on an individual’s circumstances, behaviour or choices.

    In extreme cases, it might exclude or discriminate against individuals. Decisions that might have little impact generally could have a significant effect for more vulnerable individuals, such as children.

    Example

    A social security process which automatically evaluates whether an individual is entitled to benefit and how much to pay is a decision ‘based solely on automated processing’ for the purposes of Article 22(1).

    As well as having a legal effect, the amount of benefit received could affect a person’s livelihood or ability to buy or rent a home so this decision would also have a ‘similarly significant effect’.

    Other similarly significant effects include:

    • automatic refusal of an online credit application; or
    • e-recruiting practices without human intervention.

    Example

    An individual applies for a loan online. The website uses algorithms and automated credit searching to provide an immediate yes/no decision on the application.

    Example

    As part of their recruitment process, an organisation decides to interview certain people based entirely on the results achieved in an online aptitude test. This decision has a significant effect, since it determines whether or not someone can be considered for the job.

    By contrast, the following example is unlikely to have a significant effect on an individual:

    Example

    Recommendations for new television programmes based on an individual’s previous viewing habits.

    If you are unsure whether a decision has a similarly significant effect on someone you should consider the extent to which it might affect, for example, their:

    • financial circumstances;
    • health;
    • reputation;
    • employment opportunities;
    • behaviour; or
    • choices.

    The guidelines produced by WP29 include more advice on identifying legal or similarly significant effects.

    Used correctly, automated decision-making is useful for many businesses. It can help you to interpret policies correctly and make decisions fairly and consistently.

    The GDPR recognises this and doesn’t prevent you from carrying out profiling or using automated systems to make decisions about individuals unless the processing meets the definition in Article 22(1), in which case you’ll need to ensure it’s covered by one of the exceptions in Article 22(2). See 'When can we carry out this type of processing' for more information about the exceptions.

    Creating or applying a profile to someone in order to tailor your marketing is a decision about them, but for profiling or automated decision-making to be restricted by Article 22 there needs to be:

    • no human involvement; and
    • a legal or similarly significant effect on the individual.

    The GDPR isn’t designed to stop you from running your business or promoting your products and services. However, there could be situations in which marketing may have a significant effect on the recipients. You need to think about your target market. For example, vulnerable groups of individuals may be more easily influenced and affected by behavioural advertising.

    The GDPR highlights that children in particular deserve additional protection, especially where their personal information is used for the purposes of marketing and creating online profiles.

    “Children merit specific protection with regard to their personal data, as they may be less aware of the risks, consequences and safeguards concerned and their rights in relation to the processing of personal data. Such specific protection should, in particular, apply to the use of personal data of children for the purposes of marketing or creating personality or user profiles...”

    Recital 38

    Also remember that people can object to you profiling them for marketing purposes under Article 21. You must bring details of this right to their attention and present it separately from other information. If you receive this type of objection you must stop the processing and confirm to the individual that you have done so within a month of receipt.

  • When can we carry out this type of processing?

    In detail

    You can only carry out this type of processing if you can rely on one of the three exceptions set out in Article 22(2). These exceptions are not the same as the lawful bases for processing required under Article 6.

    • When the decision is necessary for a contract

    You need to consider whether the decision is actually necessary for the performance of a contract with the individual. This means that it does not have to be essential, but it should be a targeted and reasonable way of meeting your contractual obligations. It should also be the least privacy intrusive way to reasonably achieve your objective.

    The wording of this exception implies that the decision-making defined in Article 22(1) could potentially be carried out by a different controller than the one who is party to the contract with the individual.

    “if the decision is necessary for entering into, or performance of, a contract between the data subject and a (not the) data controller.”

    Article 22(2) (a)

    Example

    A loan application might represent a contract between a financial organisation and a potential borrower. The financial organisation relies on an automatically generated credit score carried out by a credit reference agency to decide whether or not to agree the loan.

    Even though the contract is not between the data subject and the credit reference agency the decision is covered by Article 22(2)(a) as long as it can be shown that it is necessary for the contract to be fulfilled.

    • When the decision is authorised by law.

    The decision has to be authorised by law, but this doesn’t mean that there has to be a law which explicitly states that solely automated decision-making is authorised for a particular purpose. The Data Protection Act 2018 (DPA 2018) refers only to a decision which is ‘required or authorised by law’ (Chapter 2, Part 2, Section 14 (3)(b)).

    If you have a statutory or common law power to do something, and automated decision-making/profiling is the most appropriate way to achieve your purpose, then you may be able to justify this type of processing as authorised by law and rely on Article 22(2)(b). However you must be able to show that it’s reasonable to do so in all the circumstances.

    Example

    In the financial services sector, an organisation might use automated decision-making, including profiling, to identify fraud, in order to comply with a high level regulatory requirement to detect and prevent crime. It identifies cases of potential fraud by comparing data from credit reference agencies, bank accounts, the Land Registry, the DVLA, credit card sales, online marketplaces and social media.

    • When the decision is based on the individual’s explicit consent.

    Firstly you need to understand what explicit consent looks like. Consent generally under the GDPR must be a freely given, specific, informed and unambiguous affirmative indication of the individual’s wishes.

    Explicit consent means that the individual should expressly confirm their consent, for example by a written statement, filling in an electronic form or sending an email. Our guidance on consent provides more information on this area.

    In the context of Article 22, in order to be specific and informed your consent request needs to explain that the decision will be entirely automated.

    Article 22(4) provides an additional layer of protection for special category personal data. You can only carry out the processing described in Article 22(1) if one of the above exceptions applies and:

    • you have the individual’s explicit consent; or
    • the processing is necessary for reasons of substantial public interest. Substantial public interest conditions are set out in Schedule 1 Part 2 of the Data Protection Bill.
    Further reading

    Consent guidance

    The Article 29 Working Party includes representatives from the data protection authorities of each EU member state. It adopts guidelines for complying with the requirements of the GDPR

    WP29 guidelines on automated individual decision-making and profiling – Chapter IV

    WP29 guidelines on consent

  • What else do we need to consider if Article 22 applies?

    In detail

    This type of processing is high risk so you need to carry out a DPIA and introduce other appropriate safeguards, such as providing specific information to individuals about the processing and the rights available to them.

    These safeguards and rights are highlighted in the GDPR and explained in more detail in the WP29 Guidelines on automated decision-making and profiling.

    A DPIA is a tool to help you assess the risks to individuals from a processing operation and identify ways to address those risks. They are mandatory for processing that is likely to result in a high risk to individuals. For information about the types of processing requiring a DPIA please read our guide to the GDPR.

    A DPIA can help you decide whether or not the intended processing is going to be subject to the provisions of Article 22. If you already know that this is the case you must carry out a DPIA.

    Even if Article 22 doesn’t apply (because the processing isn’t solely automated), you are still required to carry out a DPIA if the processing constitutes:

    “a systematic and extensive evaluation of personal aspects relating to natural persons which is based on automated processing, including profiling, and on which decisions are based that produce legal effects concerning the natural person or similarly significantly affect the natural person”

    Article 35(3)(a)

    It’s also a good way to meet your accountability obligations by showing that you have:

    • considered the risks involved in any profiling or automated individual decision-making process; and
    • put procedures in place to mitigate those risks and comply with the GDPR requirements.

    You must inform individuals if you are using their data for solely automated decision-making processes with legal or similarly significant effects. This applies whether you have received the data directly from the individuals concerned or from another source.

    You must also provide meaningful information about the logic involved and what the likely consequences are for individuals.

    This type of processing can be invisible to individuals so in circumstances where it can have a significant impact on them you need to make sure they understand what’s involved, why you use these methods and the likely results.

    Providing ‘meaningful information about the logic’ and ‘the significance and envisaged consequences’ of a process doesn’t mean you have to confuse people with over-complex explanations of algorithms. You should focus on describing:

    • the type of information you collect or use in creating the profile or making the automated decision;
    • why this information is relevant; and
    • what the likely impact is going to be/how it’s likely to affect them.

    Example

    An on-line retailer uses automated processes to decide whether or not to offer credit terms for purchases. These processes use information about previous purchase history with the same retailer and information held by the credit reference agencies, to provide a credit score for an online buyer.

    The retailer explains that the buyer’s past behaviour and account transaction history indicates the most appropriate payment mechanism for the individual and the retailer.

    Depending upon the score customers may be offered credit terms or have to pay upfront for their purchases.

    You must provide specific information if you carry out automated decision-making described in Article 22(1), namely information on:

    “the existence of automated decision-making, including profiling, referred to in Article 22(1) and (4) and, at least in those cases, meaningful information about the logic involved, as well as the significance and envisaged consequences of such processing for the data subject.”

    Articles 13(2)(f) and 14(2)(g)

    Our guidance on the right to be informed explains the different techniques you can use, but it is good practice to use the same medium you use to collect personal data to deliver privacy information.

    If you’re operating in an online environment it shouldn’t be too difficult. Layered privacy notices or just-in-time notifications that tell people what you’re going to do with their data at the point you collect it are good approaches to adopt.

    Privacy dashboards that that inform people how their data is used, allow access to the information you hold on them, including details of any profiles and the data input into them, and allow them to manage what happens with it are other useful tools.

    If you plan to use personal data for any new purposes, you must update your privacy information and proactively bring any changes to people’s attention.

    If someone is unhappy with a decision you’ve made using a solely automated process, they can ask for a review. It makes sense for you to explain how they can do this at the point you provide the decision.

    You need to show how and why you reached the decision, so it’s important that you understand the underlying business rules that apply to automated decision-making techniques.

    You must be able to verify the results and provide a simple explanation for the rationale behind the decision.

    The systems you use should be able to deliver an audit trail showing the key decision points that formed the basis for the decision. If your system considered any alternative decisions you need to understand why these were not preferred.

    You should have a process in place for individuals to challenge or appeal a decision, and the grounds on which they can make an appeal. You should also ensure that any review is carried out by someone who is suitably qualified and authorised to change the decision.

    The reviewer should take into consideration the original facts on which the decision was based as well as any additional evidence the individual can provide to support their challenge.

    Example

    Your request/application has been declined.

    We have made this decision using an automated scoring system that takes into account the information you provided as well as...

    If you wish to discuss or appeal the decision then please contact us using the attached form. You can send this electronically or by post.

    Our review panel will consider your request and contact you within xxx days with our findings.

    Please attach any evidence that you believe might help your appeal. Examples of the types of information that may be useful in the review process are detailed below…

    Information about...

    Copies of...

    Documentary evidence of...

    You must respond to the individual within one month of any challenge or appeal. If your decision-making is authorised by law, you should check the specific requirements around response times in the latest version of the Data Protection Bill.

    Individuals have a right of access to the same details about automated decision-making that you must provide under your privacy information. If someone submits an access request Article 15 says that you have to tell them about:

    “the existence of automated decision-making, including profiling, referred to in Article 22(1) and (4) and, at least in those cases, meaningful information about the logic involved, as well as the significance and envisaged consequences of such processing for the data subject.”

    Article 15(1)(h)

    The GDPR says that where possible you should be able to provide remote access to a secure system that provides individuals with direct access to their personal data. This is also a good way for individuals to verify and check that the data you are using is accurate.

    Example

    An organisation that uses profiling in order to determine insurance premiums allows its customers to inspect and correct any inaccuracies in the personal data used in the profiles assigned to them. As well as addressing any errors, this also helps improve the precision of the system used to carry out the processing.

    You need to make sure that you have mechanisms in place to diagnose any quality issues or errors and a process to document how these are resolved.

    These mechanisms should also allow you to check that your systems are working as intended and highlight any inaccuracies or bias.

    Consider how you can:

    • introduce sample quality checks on the results from your systems to remove any bias and/or discriminatory effects;
    • ensure any special category data that may have been inferred by the profiling is deleted if it is not required;
    • identify appropriate retention policies for the information you use and keep these under review;
    • implement suitable security measures such as access controls and encryption; and
    • audit your machine-learning tools to check for decision-making rationale and consistency.
    Further reading

    Right to be informed

    Big data, artificial intelligence, machine learning and data protection (external link)

    The Article 29 Working Party includes representatives from the data protection authorities of each EU member state. It adopts guidelines for complying with the requirements of the GDPR

    WP29 guidelines on automated individual decision-making and profiling Chapter IV and VI

    WP29 guidelines on Data Protection Impact Assessment

    WP29 guidelines on transparency

  • What if Article 22 doesn’t apply to our processing?

    In detail

    Even if Article 22 does not apply to your processing, for example because there is human involvement in the decision-making process, you must still comply with the GDPR principles and identify and record your lawful basis for the processing. This is also the case for any new profiles you create.

    You may want to revise your privacy policy and inform individuals about any automated decision-making involving profiling that you carry out, especially if it’s unlikely they would expect it. You should still tell people what data you’re using and where it’s come from.

    Even if you don’t think the processing falls within Article 22, it’s good practice to do this and helps you be more transparent – particularly because this type of processing won’t necessarily be obvious to individuals.

    As part of assessing whether your processing is fair, you also need to think about whether any profiling is fundamentally linked with the reason for using the service and what people would reasonably expect you to do with their data. Although a retailer analysing loyalty card data to decide what new products to suggest to a customer might be an expected activity; analysing someone’s social media posts to reject them for a job, refuse a loan, or increase their insurance might not be.

    Individuals have access rights to the personal information you process about them, including information used in any profiling you carry out.

    Remember as well that if the data you’re using isn’t correct then any profile or decision based on the data will also be flawed.

    Don’t collect too much information or keep it for too long. Just because your systems allow you to retain vast quantities of data doesn’t mean you should. It also makes it more difficult to keep the data up to date, accurate and relevant for any profiling you’re carrying out.

    Article 21 of the GDPR gives individuals the right to object to any profiling that you carry out:

    • on the basis of legitimate interests or on the basis of public task or official authority. In these cases an individual can object on grounds relating to his or her particular situation. You’ll have to stop the processing unless you can show that you have a compelling reason to continue the profiling that overrides the individual’s interests;
    • for direct marketing purposes. You must stop the profiling as soon as you receive an objection. There are no exemptions or grounds to refuse.

    You must bring this right to the attention of individuals and present it separately from other information.

    If you receive an objection under Article 21 you need to respond within one month and confirm the action you’ve taken.

    Further reading

    Guide to the GDPR – Principles, Lawful bases for processing, Individuals rights

    Guidance on Children and the GDPR

    The Article 29 Working Party (WP29) includes representatives from the data protection authorities of each EU member state. It adopts guidelines for complying with the requirements of the GDPR.

    WP29 Guidelines on Automated individual decision-making and profiling Chapters III and V

    WP29 Guidelines on consent

    WP29 Guidelines on transparency

No questions matching current filter

Thank you for reading.

Was this article helpful?
Dislike 0
Views: 70