SARmaggedon Days Are Here Again (Again)

Reading my emails, a headline leapt out at me: “The hidden cost of GDPR data access requests“. It led me to BetaNews, a website that looks like it is trapped in 1998, and a story describing research into SARs commissioned by Guardum, a purveyor of subject access request handling software. A sample of 100 Data Protection Officers were consulted, and you’ll never guess what the research uncovered.

SARs, it turns out, are time consuming and expensive. I award 10 GDPR points to the Guardum CTO for knowing that SARs weren’t introduced in 2018, but I have to take them away immediately because he goes on to claim that “There has also been a marked change in the way that lawyers are using DSARs as part of the data discovery process.” Apparently, lawyers are using SARs now. Imagine that. The article goes to say that “Fulfilling DSARs can involve finding, compiling and redacting data in digital and paper format across multiple departments both on company networks and in the cloud.“. There’s also a bit of a spoiler about whether the Pope is a Catholic.

According to Guardum, the average cost of a SAR is £4,884.53, the average DPO receives 27 SARs a month, and each one takes an average of 66 working hours to deal with. The article didn’t explain how these figures were arrived at, so I eagerly clicked the link to visit Guardum’s website for the full results. What I found was a fountain of guff. Strip out the endless bar and pie charts, and what Guardum wants to say is that 45% of the DPOs surveyed would like to automate some of the process because of a predicted landslide of SARs, provoked by angry furloughed and sacked staff.

I’m not sure about the logic of this – I can understand that everyone who loses their job will be upset and probably angry, and I’ve certainly dealt with lots of SARs related to a suspension or dismissal. But in those cases, the action taken was personal and direct – an individual was singled out by the employer for the treatment in question. I don’t see why people losing jobs in a pandemic will be so determined to send a SAR. It’s not like the reason for their predicament is a mystery.

The survey questions are opportunistic at best, and at worst, seem designed to allow Guardum to paint this picture of anxious DPOs uncertain about how they’re going to handle the post Covid-19 SARmageddon that the company is evidently desperate for. 75% of respondents are described as having difficulties dealing with SARs during the lockdown, though this actually translates as good news. 72% are coping but expect a SAR backlog when they get back to the office, while just 3% fearing a ‘mountain’ of requests. The headline on one slide is that 30% anticipate a ‘massive’ increase in SARs, but the reality is 55% expect the same as before and 15% think they’ll get less. 73% supposedly think that furloughed or laid off staff will be a ‘big factor’ in the predicted increase, even though the breakdown shows that only 20% think it will be the single biggest factor. To emphasise, these are requests that haven’t happened yet. The people who say that they will are the ones flogging the software to deal with the problem.

So far, so what? Guardum have software to sell and a cynical pitch about Covid-19 to achieve that. Does it matter? In the grand scheme of things, no, it doesn’t. I’m probably not the only person currently experiencing a crash course in What’s Really Important. But in the micro scheme of things, bullshit deserves to be called out, especially when it’s designed to exploit a crisis that’s causing misery and death across the world. Many of the revelations in this survey are staggeringly banal – nearly 50% of people find tracking the data down across multiple departments to be a slog, while 63% have to search both paper and electronic records. Who with any experience in Data Protection would think it was worth pointing this out? Meanwhile, the assertions about how long a SAR takes or how much it costs are wholly unexplained. It’s meaningless to claim that the mean cost of a SAR is £4,884.53 if you don’t explain how that was calculated (inevitably, the CTO is touting this figure on LinkedIn).

Guardum aren’t necessarily the experts at Data Protection that they might have us believe. For one thing, despite being a UK company, both the survey results and their website exclusively refer to ‘PII’ rather than personal data. For another, part of the criteria for participating in the survey was that the DPO needed to work for a company with more than 250 employees. This was, for a time, the threshold for a mandatory DPO but despite being changed, some dodgy training companies and consultants didn’t notice and ran courses which highlighted the 250 figure even when it was gone. Most importantly, nearly half of the people who responded to the survey don’t know what they’re doing. The survey was purportedly targeted at DPOs, but 44% of respondents are identified as being in ‘C-level’ jobs – perhaps this is to give a veneer of seniority, but C-level jobs are precisely the senior roles that are likely attract a conflict of interests. Guardum talked to people in the wrong jobs, and apparently didn’t realise this.

The ‘About’ page of Guardum’s website proclaims “Guardum supports privacy by design – where data privacy is engineered into your business processes during design rather than as an afterthought“, but the execution is less confident. There is a questionnaire that shows how much an organisation can save by using the Guardum product, but when you complete it, you have to fill in your name, company and email to get the results, and there’s no privacy policy or transparency information about how this information will be used. Moreover, if you try to use the contact form, clicking on the link to the terms and conditions results in ‘page not found’.

I have to declare my bias here – I don’t believe that any ‘solution’ can fully deal with the SAR response process, and I think people who tout AI gizmos that automatically redact “PII” are probably selling snake oil. Some of the SAR grind comes in finding the data, but a lot of it is about judgement – what should you redact? How much should you redact? Anyone who claims that they can replace humans when dealing with an HR, mental health or social care is writing cheques that no product I have ever seen can cash. So when I land on a website like Guardum’s, my back is up and my scepticism is turned all the way up. It would be nice if once, I saw a product that wasn’t sold with bullshit. But not only is Guardum’s pitch heavy with management buzzwords, they’re using fear as a marketing tool. Just last week, they ran a webinar about weathering the ‘Post Pandemic DSAR Storm‘.

Guardum claim that they provide “the only solution that can fully meet the DSAR challenge of responding in the tight 30-day deadline, giving you back control, time and money that are lost using other solutions“. Nowhere do they mention that you can extend the deadline by up to two months is a request is complex (and many are). But even if their claims are true, why do they need to sell their product via catastrophising? If their expertise goes back to the 1984 Act, why are they calling it PII and talking up the opinions of DPOs who are in the wrong job? Why oversell the results of their survey? Why hide the basis of the hours and cost calculations on which is all of this is being flogged?  And what on earth is a ‘Certified Blockchain Expert‘?

The future post-Covid is an uncertain place. I find the utopianism of some commentators hard to swallow, partly because people are still dying and partly because the much-predicted end of the office will have career-changing consequences for people like me. But at least the LinkedIn prophets are trying to explore positives for themselves and others in an undeniably grim situation. The people running Guardum seem only to want scare people into getting a demo of their software. If one is looking for positives, the fact that the ICO has waved the white flag means that no organisation needs to be unduly concerned about DP fines at the moment, and despite some of the concerns expressed in Guardum’s survey, nobody in the UK has ever been fined for not answering a SAR on time. The old advice about deleting data you don’t need and telling your managers not to slag people off in emails and texts will save you as much SAR misery as any software package, and I can give you that for free.

“masterclass in not answering questions”

Just about a month ago, I had a little Twitter disagreement with Paul-Olivier Dehaye, patron saint of subject access requests. He said his tool for making subject access was brilliant and revolutionary, and I said it was shit. There was a bit more to it than that, but I was hoping to make this a short blog.

The use of third parties to make subject access requests on one’s behalf is not new – solicitors have always done it, and companies have made batched SARs at least since the bank charges furore of the last decade. The problem with a third party – or automation of the process – is that it gives the Data Controller something to play with. Dehaye admitted to me that in all the time he spent developing his SAR tool, he didn’t speak to anyone with any experience of dealing with SARs from the controller’s perspective, and it shows.

Even though one of Dehaye’s tedious cheerleaders told me that SARs were going to be “frictionless” post-GDPR, there are inevitably some bumps in the road when asking for data even in this Brave New World. The Data Controller needs to identify the application properly, and the involvement of a third party might complicate that – or might be exploited to complicate that, as anyone who has ever dealt with a poorly-written solicitor SAR can probably tell you. If there is a lot of data, the controller can ask the subject to narrow the scope of their request. If they believe that the request is unfounded or excessive, they can make a charge, or even refuse. An automated third party doesn’t make any of this easier.

Ironically given his status as pro-DP activist, I think Dehaye wants SARs to seem difficult. “In my own experience, SARs are complicated to do in a way that properly defends data subject rights” he said, but given that he’s building a business based on data, he kind of would say that. When I first encountered him, Dehaye told me that he was planning to charge subjects for using his tool; while that plan might have changed, he gets evasive when you ask whether he might charge for add-on services in the future. One of the main advantages of GDPR for the subject is that SARs are now free – the best way to exercise the right is to ask for the data direct, without the involvement of a politically-motivated middleman whose company isn’t even in the EU. I voted Remain and I think Brexit is moronic, but that doesn’t mean that weaponising SARs is a good idea. After all, someone might turn round and do it to you.

I decided to make a SAR to Dehaye’s company on the 25th May. His response, though admirably swift, wasn’t exactly the zenith of transparency that one might have hoped for. One might even describe it as a masterclass in not answering questions. I provided a variety of different email addresses and phone numbers that the company might hold in relation to me – the purpose of this was to allow the data controller to identify whether any of my data was held. I did the same thing with my request to Experian – I don’t know what data Experian holds on me, so I provided all the possible identifiers that I could think of. I don’t know what, if any, data Dehaye or his company might hold, so I needed to provide a variety of different identifiers.

EDIT: in response to a request from the data controller, click here for the full text of my request (redacted only to remove personal data that is not in the public domain) and the full text of their reply.

Article 12 of GDPR states that “The controller shall facilitate the exercise of data subject rights under Articles 15 to 22” and shall answer requests unless it “demonstrates that it is not in a position to identify the data subject” – it is plainly correct for the controller to want to know who the applicant is, in order to avoid giving data to the wrong person. However, Recital 64 says that the controller’s measures to identify the subject must be “reasonable“. Dehaye demanded that I send a separate request from each of the email addresses I specified. This means that he thinks that if an organisation has harvested emails from a variety of sources, the controller only has to disclose data if they receive confirmation from that account that it is linked to the subject. So if a person applies from a Gmail account, and the controller has harvested a work email address, even if they have linked the two together, Dehaye doesn’t think that the subject is entitled to the work-related data unless they make a separate request.

Similarly, I provided my home address, my 2 mobile numbers (business and personal) and my landline. Bear in mind, a data controller may have harvested all of this data, so the SAR applicant might need to provide it in order to say this is me, this is my data, do you have it? Dehaye’s response to this part of my request was to demand copies of phone bills for each account, and a recent utility bill for the home address. Clearly, this is the approach he would advocate for any data controller faced with such a request. As it happens, my girlfriend’s name is on the landline account, so I cannot prove that the landline is my personal data, even though it is. One of my mobiles is pay-as-you-go, so I don’t get bills, and the work mobile is on my website, and so can be linked to me without the need for unnecessary proof. As with most people, I receive electronic utility bills, and do not have them immediately to hand. Dehaye’s approach seems to be that if a Data Controller has harvested your data, subject access requires the applicant to provide a lot more personal data in order to get access.

The point of the ID check is to ensure that the person is who they say they are – once that’s done, if the controller has doubts about whether an identifier does link back to the subject (i.e. an email address), they can check, or just send any relevant data to that separate identifier. If Dehaye thinks that his approach is legally correct, there is no reason why Leave.EU, Vote Leave or any other organisation shouldn’t do exactly the same thing if they receive a SAR from now on. When I asked him in April how his tool would deal with the ID element he said “Let’s set the standard” – now we know what that looks like. It looks like giving huge quantities of personal data to someone you don’t trust.

This is a no-win – either Dehaye’s approach is right, and I have to go through an administrative nightmare when SAR-ing organisations that grab data from anywhere they can get it, providing them with a fat dossier of extra information before I can get access, or Dehaye is a hypocrite who complains about hurdles to subject access but builds a wall when asked to practice what he preaches. In any case, if Dehaye’s obstructive and unhelpful approach was correct, it would still be easier to handle without the added complication of a middleman.

UPDATE 28/5/18: Mr Dehaye has admitted that he deliberately adopted an obstructive approach because he thinks I am a trouble-maker. I believe that this is a clear breach of the GDPR; if the Data Controller Personal Data.IO is capable of playing these kinds of games, and deliberately discriminates against data subjects, I think this seriously undermines their credibility to act as an agent for other people’s SARS. The company is setting a cynical, obstructive example, and it would be catastrophic for subject rights if other controllers followed their lead.

The Purge

Throughout the campaign for the Labour leadership, various people applying to be registered supporters have had their applications rejected. The list is varied, from the film director Ken Loach and the comedian Mark Steel, through to the human equivalent of genital herpes, Toby Young. Those registering to be supporters must agree that they support the aims and ideals of the Labour Party: Loach and Steel have explicitly and recently advocated voting for other parties, while Young is a high-profile Conservative. I’m not going to lose any sleep, frankly. However, in the past couple of days, a substantial number of less well-known people have received similar missives – some were recently candidates for other parties so Labour’s ban may have some merit. But others are just ordinary people on the left. Some of them are critics of austerity, some may have said that they are voting for the Greens or the Trades Union and Socialist Coalition, or just slagged off the Labour leadership online. I think Tony Blair is a war criminal and have said so often, so I still wonder if my vote yesterday counts. Is that acceptable for Labour High Command in the current climate?

The Data Protection problem for Labour is that when we signed up to be registered supporters, there was no clear fair processing information explaining that we would be vetted or how this would be done. Some form of vetting has clearly happened – I’ve even seen copies of emails and Facebook posts that suggest a full-on witch-hunt for anyone who isn’t an uncritical supporter of the party. I’m not sure whether these are real, but there are a lot of them.

As I have previously written, Labour does not need consent to look at websites and Twitter accounts. Even though the stuff on Twitter is sensitive personal data as it relates to political opinions, Data Protection allows for sensitive data to be used if it has been put into the public domain by the data subject. Furthermore, I agree that Labour has a legitimate interest in preventing full-on Tories from voting. This means that they can rely on the ‘legitimate interest’ justification to use personal data. However, they are required not to cause unwarranted prejudice to the rights and freedoms of data subjects when doing so. This is all part of the first Data Protection principle. I believe that legitimate interest requires the vetting process to be carried out objectively and accurately. Without some form of appeal, I think the rights and freedoms of the data subjects have been prejudiced.

More fundamentally, Labour must also process data fairly. The blurb for registered supporters was thin, so as someone who signed up, I have no idea what process was gone through. Even if you are one of those (wrong) people who thinks that trawling Twitter doesn’t engage Data Protection, receiving and acting on tip-offs and reports isn’t just disturbingly McCarthyite, it would be a breach of the Data Protection Act unless registered supporters were told. There are in fact a host of potential problems (accuracy, relevance, security), but the fairness one is enough because it is insurmountable. We should have been told – we weren’t.

Even if you think such a process would be legitimate, there is no exemption from the Data Protection Act, nothing that allows Labour to do these things secretly. The exemptions in Data Protection cover legal proceedings, criminal investigations, cases referred to regulators – situations where handling personal data secretly can be justified. None of the exemptions applies to the kind of process currently at work in the Labour Party. The foundation stone of Data Protection is fairness and transparency – letting people know how their data is used, so that they can ensure it is used properly. Not for the first time, the Labour Party is acting secretively, and so I have not faith in the vetting process. I suspect it breaches the first Data Protection principle.

Data Protection gives every person a right of subject access, a right to request copies of their personal data held by any organisation. In this case, the data on which the decision was made to ban a person from voting in the leadership election will undoubtedly be personal data. Admittedly, Labour could claim that no data was recorded, but this would reveal that process to be slapdash in the extreme.

Therefore, my advice to anyone rejected by the Labour Party is as follows: make a subject access request. Find out what it was that made Labour reject you, and then publicise that. Expose this process, and dig it over. Labour did not want this to be a transparent process, but they cannot stop you from finding out what happened in your case.

To make a subject access request, you need three things:

  1. A written request, setting out your name, address and the email address you registered with as a supporter
  2. Proof of your ID. Send a copy of a passport or driving license and ask them to destroy it when they have validated your request. They can refuse to deal with your request without proof of ID, so don’t give them the opportunity to delay by asking for it
  3. A cheque for £10. Having already lost the £3 supporter fee, this will be annoying, but I doubt Labour will accept a subject access request without the statutory fee, and they can refuse to process the request without it. If you want to know what happened (or find out that it was a flawed process), you will have to sacrifice the tenner. If they are feeling generous, they won’t cash the cheque. The Information Commissioner cannot order them to waive the fee, so don’t waste your time asking them.

You may well want to send this by recorded or registered post, which ratchets up your costs even more. If you are throwing your hands up in despair at spending another £12, I’m sorry. I didn’t say you would like my advice. Explain clearly and simply that you want all of the personal data held about you as a registered supporter, including any and all information that was used to ban you from voting. You are entitled to a permanent copy of the data. It is unlikely they will tell you the names of those involved in the decision, but the reasons you have been banned must be made available. It doesn’t matter of hundreds of Labour supporters make a subject access request at once – there is no provision to refuse vexatious requests, and the Information Commissioner’s Code of Practice on Subject Access makes clear to organisations that they must be prepared to respond to peaks in demand.

Subject Access is an imperfect tool: organisations sometimes don’t record the information you expect them to. But Labour took their supporters’ money and then denied many of them a vote. Either they have to account for these decisions, or admit that they have not done so fairly. Those calling for the election to be halted to avoid a Corbyn victory should be full-throated in their demands that the banned should either get a proper explanation as required by Data Protection, or the vote should instead be halted until a proper process is undergone.

The address Labour publish to contact their Data Protection Officer is

Compliance Unit, Labour Party, One Brewer’s Green, London, SW1H 0RH

If you go for it, good luck. Drop me a line and let me know how you get on.