Posts on Chris Smith - Privacy Engineer

Cyber-security incident involving PHP and responsible disclosure practices

Thu, 22 Oct 2026 00:00:00 +0100

Puzzling to me that researchers operating in cyber-security appear not to have followed responsible disclosure practices.

Seems certain details of the vulnerability were publicly disclosed after researchers “mistakenly published a video” [1] and “details of the flaw leaked on social media on Tuesday” [2]. PHP Foundation had intended to release details of the vulnerability, along with a complete fix, later this month [2].

Interesting to consider the extent to which researchers and their organisations can, and should, be held responsible for the disruption caused by non-adherence to responsible disclosure practices.

New article on the relationship between anonymisation and anonymous information

Wed, 14 Oct 2026 00:00:00 +0100

The relationship between ‘anonymisation’ and ‘anonymous information’ is not straightforward in theory or practice.

I explored the relationship between these two concepts using recent draft guidance on anonymisation and pseudonymisation from the Information Commissioner’s Office.

One area where I feel further clarification would be helpful is how ‘anonymous information’ can be simultaneously:

Not subject to UK GDPR
Contingent on the party processing the information (‘whose hands’)

Read the article at: https://www.drcjsmith.com/articles/exploring-the-relationship-between-anonymisation-and-anonymous-information/

Microsoft President's comments on AI safety and regulation

Mon, 12 Oct 2026 00:00:00 +0100

Some sensible thoughts on AI safety from Microsoft’s President, including a ’legal obligation to tell a “victim” if their system is breached by a rogue AI agent'.

Seems very problematic to have AI systems being developed, procured and used in contexts where legal responsibility and systems of accountability for their decisions and actions are under-defined.

Read the BBC article describing the comments at: https://www.bbc.co.uk/news/articles/c6rmy7zmvx3lo

Irish Data Protection Commission's decision to issue a €403 million fine to Google

Thu, 08 Oct 2026 00:00:00 +0100

The Irish Data Protection Commission’s decision to issue a €403 million fine to Google highlights the importance of accountability in privacy and data protection.

The DPC cited four specific infringements of the GDPR by Google, including an infringement relating to its accountability obligations under Article 5(2), by “failing to demonstrate compliance with the lawfulness, fairness and transparency principle regarding its processing of personal data in Location Accuracy” [DPC Decision].

As described in Google documentation, the Location Accuracy feature uses available wireless signals, including GPS, WiFi and Mobile networks; and available sensors, including accelerometer, barometer and gyroscope to determine location.

Privacy International's response to the DSIT's call for evidence relating to data regulation

Thu, 01 Oct 2026 00:00:00 +0100

Privacy International’s response to the Department for Science, Innovation & Technology’s call for evidence highlights how AI challenges fundamental tenets of data protection law.

In particular, that organisations who are responsible for processing are able to define what, how and why personal data are being processed at all stages of a system’s lifecycle.

Due to the potential complexity and opacity of processing associated with the different stages of an AI system’s lifecycle, organisations who are responsible for these systems may be unable to comprehensively characterise the personal data that they process.

OpenAI agent's 'infiltration' of an Australian government website

Thu, 24 Sep 2026 00:00:00 +0100

A couple of brief thoughts regarding OpenAI agent’s ‘infiltration’ of an Australian government website:

In terms of responsibility, AI agents can be reasonably characterised as acting ‘on behalf of’ a particular individual or organisation. Therefore, the decisions and actions taken by an AI agent, and the conformance (or not) of these actions with law, norms and cultures can be reasonably attributed to that individual or organisation. If a human agent acting on behalf of an organisation had taken the same decisions and actions as the AI agent then attribution of responsibility would seem straightforward.

Initial version of AI Usage Notice published

Tue, 22 Sep 2026 00:00:00 +0100

An initial version of our AI Usage Notice is now available.

The notice provides information about our approach to the use of Artificial Intelligence (AI) systems, the specific AI systems that we use, and the different activities for which we use these systems.

As outlined in the notice, we are committed to the trustworthy use of AI systems. We will always strive to act with honesty, competence and integrity in our usage of AI systems.

Joint Committee report on Human Rights and the Regulation of AI

Mon, 14 Sep 2026 00:00:00 +0100

“AI systems pose particular risks to privacy and personal data….This threatens the right of privacy of people in the UK and elsewhere” (Paragraph 45).

Joint Committee report on Human Rights and the Regulation of AI also highlights the complexity of balancing privacy with other rights and interests, and the difficulties faced by different actors in the AI lifecycle who are/will be required to make these complex balancing decisions when developing and deploying AI systems (see Paragraph 40).

National Commission into the Regulation of AI in Healthcare

Thu, 10 Sep 2026 00:00:00 +0100

Privacy and data protection are likely to be key determinants of public trust in AI and the application of AI in healthcare.

Indeed, previous problematic data initiatives in the NHS have highlighted the importance of privacy and data protection and the key role it plays in public trust in healthcare. For example, data sharing by the Royal Free London NHS Foundation Trust with Google Deepmind.

Any regulatory framework for AI in healthcare therefore needs to place privacy and data protection at its core.

Mozilla Foundation investigation into Period and Ovulation Apps

Mon, 03 Aug 2026 00:00:00 +0100

Mozilla Foundation investigation shows how important it is for app developers to fully understand the actual data practices that result from their technical and business decisions.

Without this understanding, app developers cannot meaningfully:

Determine whether these practices are consistent with their own expectations, and the expectations of users, regulators and other stakeholder groups.
Substantiate any privacy claims relating to these practices that are made within privacy notices and marketing materials

Privacy International investigation into AI recruitment platforms

Thu, 23 Jul 2026 00:00:00 +0100

Very interesting investigation into AI recruitment platforms from Privacy International.

Report highlights some significant issues in the platforms tested, including:

Different ‘match scores’ awarded to identical (fictional) candidates with the same CVs for identical (fictional) job openings
Higher ‘match scores’ awarded to ‘Jargon CVs’ generated by ChatGPT than CVs produced by real people, even when the ‘Jargon CVs’ contain obvious errors
Absence of meaningful explanation on how ‘match scores’ were calculated from CVs

As the report notes, these issues have implications for the reliability and fairness of recruitment processes underpinned by these platforms.

EDPB guidelines on 'web scraping' for Generative AI

Thu, 16 Jul 2026 00:00:00 +0100

European Data Protection Board (EDPB) guidelines highlight the challenges for AI developers in complying with data protection legislation when collecting data and processing data collected through ‘web scraping’.

A couple of key takeways from the guidelines:

Compliance (and demonstrating compliance) is hard. In many contexts, it may not be feasible. Determining feasibility and ensuring compliance would require a thorough understanding of, for instance, the application context, the model lifeycle, and data protection across technical and legal dimensions.

Understanding Meta's 'Muse Image' feature

Thu, 09 Jul 2026 00:00:00 +0100

Important to remember that Meta is a digital advertising platform. Any platform features need to be viewed through this lens to understand their motivations and the implications for privacy and data protection.

Briefly, the new Muse Image feature is likely to provide Meta with additional streams of data, including inputs and outputs from AI chats, that can be used to drive:

User profiling
Creation and evaluation of advertising imagery
Development and refinement of AI models

Enabling use of public profile images as inputs to the Muse Image feature (unless users explicitly opt-out) is likely to be motivated by a desire to increase user engagement - users will create, share and react to images on the platform. This drives data collection, which drives the factors above.

Driving AI adoption with robust evaluation

Thu, 02 Jul 2026 00:00:00 +0100

Ford’s story provides a valuable case study for AI adoption and highlights a pattern that may become common across organisations:

Adopt AI systems on the expectation of significant increases in effectiveness and/or efficiency for some set of tasks
Adjust the human workforce that performed these tasks on the basis that the expectation will be fulfilled
Discover through experience that the expectation of AI systems is not fulfilled for one or more of the tasks
Adjust human workforce to ‘work alongside’ the AI systems and attempt to regain the effectiveness achieved prior to AI adoption

Rather than being driven by the expectation of AI systems, organisations need to be driven by the reality.

Pausing of Meta's employee tracking initiative

Thu, 25 Jun 2026 00:00:00 +0100

Some brief points in relation to the reported pausing of Meta’s employee tracking initiative:

Appears that pausing of the initiative was prompted by concerns over security measures, with the BBC reporting that the program was halted by Meta “after realising some of the collected data had been left potentially accessible to anyone inside the company” - see https://www.bbc.co.uk/news/articles/cq615g3z36po.
Unclear what the legal basis would be for processing of personal data as part of such an initiative. Doesn’t seem reasonable to expect participation in the initiative to be a condition of employment, and seems likely that many employees would not consent to, or reasonably expect, such data to be processed as part of their employment for the purpose of training AI models.
Employee data within the initiative may be contributing to the training of AI models that eventually automate tasks that form part (or all) of employee roles. Creates a strange dynamic where employees are potentially contributing to their own obselesence in a given role.
Employees may perform tasks within their employment that do not fall within their role description and which may reasonably be considered ‘sensitive’. For example, tasks that relate to their health and wellbeing in the workplace. Interesting to understand whether data relating to such tasks were/are/would be captured by such initiative(s), and the safeguards that would be put in place to protect such data.

Read guidance on monitoring workers from the Information Commissioner’s Office (ICO) in the United Kingdom at: https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/employment/monitoring-workers/

NCSC statement about risks posed to cyber-security by AI

Mon, 22 Jun 2026 00:00:00 +0100

National Cyber Security Centre (NCSC) statement highlights the significant and imminent risks posed to cyber-security by AI.

Security-by-design and security-by-default approaches required to understand, control and communicate these risks go hand-in-hand with data protection-by-design and data protection-by-default approaches.

Read the NCSC statement at: https://www.ncsc.gov.uk/news/the-ai-shift-in-cyber-risk-why-leaders-must-act-now.

Read the Guardian article about the NCSC statement at: https://www.theguardian.com/technology/2026/jun/22/anthropic-claude-fable-ai-model-artificial-intelligence-national-security.

How agentic coding tools will transform the software engineering profession

Thu, 18 Jun 2026 00:00:00 +0100

Agentic coding tools such as Claude Code and GitHub Co-Pilot will transform roles and careers within the software engineering profession.

Specifically, it seems likely that there will be:

Reduced demand for software engineering roles where primary responsibilities can be automated by agentic coding tools. For example, refactoring code, writing tests and authoring documentation.
Increased demand for software engineering roles where primary responsibilities cannot readily be automated by agentic coding tools. For example, technical leadership, governance and oversight.

Software engineering roles that have typically included responsibilities relating to technical leadership, governance and oversight have been those at more senior levels. However, it seems likely that such responsibilities will become expectations of a significant proportion of software engineering roles.

Apple announce access to server-based LLM via Private Cloud Compute

Thu, 11 Jun 2026 00:00:00 +0100

Developers on the App Store Small Business Program* can now get access to a server-based Large Language Model (LLM) via Private Cloud Compute (PCC) at Apple with no cloud API cost (subject to usage limits).

Based on the capabilities required for their specific use cases, developers on Apple platforms can choose whether to use Apple’s local (on-device) model, or Apple’s server model (via PCC).

Apple highlight that PCC: i) does not store data, ii) only uses data to process requests, iii) has been independently verified by researchers.

AI systems as data processing systems

Thu, 04 Jun 2026 00:00:00 +0100

Is the development, deployment and usage of AI systems in accordance with data protection requirements fundamentally any different to other data processing systems?

Whether or not a system is considered to be an ‘AI system’, organisations need to understand the (personal) data flows and data processing tasks that are associated with each stage of the system lifecycle.

Complexity of data flows and data processing tasks that comprise modern AI systems (and data processing systems in general) may increase the expertise, time and resources required to gain this understanding. However, such understanding remains essential to the fulfilment of data protection requirements at each stage in the system lifecycle.

Effects of AI dependence on trustworthiness and trust

Thu, 21 May 2026 00:00:00 +0100

Royal Observatory makes some excellent points about the potential effects of AI dependence on individuals and society. Dependence of individuals on AI tools, such as chatbots, raises some important questions regarding trustworthiness and trust. For example:

To what extent can information provided by an AI tool be considered trustworthy?
How do users determine the trustworthiness of the information provided by AI tools?
Are users making ‘good’ decisions about placing (or refusing) trust in the information provided by AI tools?

A key challenge for users in determining trustworthiness and making ‘good’ decisions is that AI tools may communicate trustworthy and untrustworthy information in the same (or similar) way(s).

Shaping the early governance of wearable technologies in society

Thu, 14 May 2026 00:00:00 +0100

Some reasons why employers and workplaces may shape the early governance of wearable technologies in society:

Employers have duties and obligations in areas that are directly impacted by use of wearable technologies in the workplace. For example, safety, equality and data protection.
Employers can identify and articulate specific harms that could arise from use of wearable technologies in the workplace. For example, harms arising from intimidation of staff, ‘chilling effects’ on organisational discourse, and breaches of confidential data.
Organisational policies can adapt quicker than legislation to the challenges posed by wearable technologies; defining expectations relating to their use (or not) in the workplace, establishing norms that are adopted more widely in society, and providing experiences that inform development of legislation and guidance.

View the BBC News article at: https://www.bbc.co.uk/news/articles/cj37z8357e5o

Governance of software development automation using AI agents

Thu, 07 May 2026 00:00:00 +0100

Automation of software development using AI agents raises important governance questions, including:

How do AI agents operationalise human values such as privacy and fairness in their processes and outputs? Human values are typically expressed using high-level principles that must be ’translated’ into practice in a given context. Given the significant challenge for humans in undertaking such ’translation’ in the design and evaluation of systems, it is unclear how (or even whether) AI agents could undertake and justify such ’translation’ with sufficient rigour for systems to be trustworthy with respect to these values
Who is responsible for the process and outputs of these AI agents? AI agents can not only generate code, but also generate the code that tests the code. Humans move further away from not only the code itself, but the assurance of the code with respect to functional and non-functional properties. Additionally, as the scope of AI agent work increases, the volume and complexity of code (and tests) generated may exceed that for which one or more humans have the capability or capacity to comprehend and reasonably oversee. Individuals within organisations may be asked to assume legal and moral responsiblity for processes and outputs that they cannot or do not comprehend, and for which their participation in assurance processes may be limited
How is the risk of supply chain attacks mitigated? AI agents may generate code that introduces dependencies on components from external package repositories. AI agents therefore require criteria on which specific repositories and packages within these repositories can be robustly evaluated to determine whether or not they should be trusted. In the absence of robust and rigorous evaluation processes, AI agents may introduce dependencies on components that are not trustworthy and place organisations at risk of supply chain attacks. Reputational and financial impacts of such attacks would negate some of the economic benefits on which use of AI agents for software development may be predicated

Read the Guardian article at: https://www.theguardian.com/technology/2026/may/07/your-craft-is-obsolete-wisetech-staff-in-limbo-as-ai-touted-as-better-than-humans

Transforming the governance of health data for research purposes

Thu, 30 Apr 2026 00:00:00 +0100

As pointed out in the article below, transformation of governance will be a key factor in determining the success of the proposed ‘Health Data Research Service’.

Some brief thoughts on ways in which the governance of health data for research purposes might be improved:

Move away from use of abstract terms such as ‘gold standard’ to describe security and privacy measures. Such terms have no meaningful semantics. Additionally, they are likely to elicit different expectations for different parties and therefore cause misunderstandings. Focus is better placed on clearly articulating the security and privacy measures that are in place, and why they are necessary and sufficient in a particular context
Harmonise the interpretation of legal frameworks and ethical principles applicable to use of health data for research purposes. Legal frameworks and ethical principles define abstract concepts and principles that must be interpreted and then operationalised in a given context. Variation in interpretation and operationalisation between the different organisations in the health data ecosystem is likely to be a source of misunderstanding and lead to ineffective and inefficient decision-making processes
Develop people and processes to make and clearly articulate informed decisions based on balancing of different (and potentially conflicting) individual, organisational and societal interests. Effective and efficient decision-making processes in the governance of health data are constrained if people and organisations do not have the expertise to confidently make and articulate these inherently complex decisions

UK Government announcement available at: https://www.gov.uk/government/news/prime-minister-turbocharges-medical-research

Anonymisation and the UK Biobank incident

Thu, 23 Apr 2026 00:00:00 +0100

Worth noting that:

‘Anonymised’ data is not ‘anonymous’. Additional technical and organisational measures are typically used in addition to ‘anonymisation’ of the data to ensure measures are adequate to protect privacy. For example, contracts and information security assurances
No ‘personally identifiable information’ does not mean no risk. Given the nature of the data items that are likely to be included in the data, it may be feasible for individuals or organisations with the willingness and ability to re-identify individuals. For example, using additional data to which they may have access
UK Biobank data involved in the incident is now potentially ‘in the wild’; simply removing from the listing(s) from a website does not mitigate the continuing risks posed by the potential existence of one or more copies of the ‘anonymised’ data outside of additional technical and organisational measures

UK Biobank update available at: https://www.ukbiobank.ac.uk/news/a-message-to-our-participants-uk-biobank-data-security-update/