This piece looks ahead to what we might expect as IT law developments in 2018.
Unusually as we go into a new year, the main headlines of what IT lawyers can expect in 2018 are signposted at the outset: new financial services laws in January, the GDPR in May and looking ahead to Brexit in March 2019. But away from the headlines, the developing narrative is more complex. Here, the combination of [mobile+social+cloud] still carries all before it and the machine learning, blockchain and digital reality of the Fourth Industrial Revolution (4IR) are quickly gaining traction whilst Moore’s Law still chugs away in the background.
What’s also fairly clear as we go into 2018, and whether it’s bitcoin, AI or tech valuations, is that we’re currently at a rather ‘frothy’ part of the tech hype cycle. As the economist J. M. Keynes famously said, ‘markets can stay irrational longer than you can stay solvent’ so there’s no telling that this will not continue through 2018. But it is worth calling to mind at the height of the dot com bubble when shares peaked in March 2000, that only 5% of the world’s population had an internet connection (it passed 50% for the first time in 2017) and that no one had then foreseen the combination of [mobile+social+cloud] that five to ten years later would be the internet’s ‘world beater’. So, if we are towards top of a bubble now, there’s precedent for thinking that we may not get it entirely right as we look out to say 2025.
Financial Services – MiFID and PSD2
Back to the present, and in January 2018, MiFID II (the Market in Financial Instruments Directive) and PSD2 (the Payment Services Directive) come into effect. MiFID II effectively takes the regulatory regime for equities trading price transparency introduced by MiFID in 2007 and extends it to trading in bonds, derivatives and most structured finance products. PSD2 takes the regulatory regime for payment services introduced by the PSD in 2009 and updates it for developments in e-commerce, the internet, mobile payments and new technologies and changing customer patterns. Both sets of rules are driven by, and in turn drive, technology change. MiFID II means rearchitecting business processes and data flows to fit the new regulatory environment and provides a great opportunity for RegTech. PSD2 provides a great opportunity for FinTech start ups. Changing financial services regulatory environments in these and other areas will continue to be an important source of work for IT lawyers in 2018.
Despite the political progress made at the end of 2017, the shape of Brexit still looks hard to call. In a 2017 report for trade association Tech UK, ‘The UK Digital Sectors After Brexit’ research consultancy Frontier Economics found that the UK’s digital sector accounted for 16% of national income, 10% of UK employment and 24% of exports. Consequently, as so much digital business from the UK is in services not goods, the primary trade-related Brexit risks for the digital sector are from services related non-tariff barriers. Accordingly, a ‘hard’ Brexit, defaulting to WTO rules based on tariff barriers for goods would not be a good option and:
“it is in the strong interests of the UK’s digitally-intensive sectors that a new comprehensive free trade agreement is reached with the EU that can enable continued growth in the UK and innovation and digital transformation across the EU.” 
Whether the calls of the large UK digital sector will get drowned out in the noise of an increasingly frazzled Brexit environment in the run up to March 2019 remains to be seen.
GDPR and data protection
GDPR will continue to be top of mind for many IT lawyers until 25 May 2018 when it enters into force and likely for a while thereafter. In the words of the Economist newspaper the GDPR ‘is “the most complex piece of legislation the EU has ever produced”’ and “will be one of the most important pieces of legislation brought into force in 2018”. As the Information Commissioner has said, ‘we’re all going to have to change how we think about data protection’ and, as everyone is finding out, you can’t just paper your way to GDPR readiness.
In addition to the rising bow wave of work in the first half of 2018, data related legal work is likely to settle back at a significantly higher level than before. GDPR regulatory enforcement will gain traction after the summer holidays and we’re likely to see over the remainder of 2018 an outbreak of litigation and the continuing weaponization of data protection claims in the employment, B2B and international contexts.
A potential curve ball to watch out for is the new ePrivacy Regulation – the source of the rules on cookies and the requirements for cookie policies. The EU published a draft regulation in January 2017 and it is still under discussion, but it looks like extending the current (2009) ePrivacy Directive significantly. Although the EU said in early 2017 that they want the new regulation to start at the same time as the GDPR, this is looking increasingly less likely, and it may be 2019 before it comes into effect.
Data governance and trust frameworks
Perhaps more significant even than this, GDPR is prompting a deeper dive across business into the governance and management of data flows within, and data sharing between, organisations. The Deepmind/Royal Free case shows what can happen when you get it wrong. Here, the ICO in July 2017 found the Royal Free Hospital had breached the Data Protection Act in sending to Deepmind unencrypted personal data of 1.6m patients for testing the hospital’s ‘Streams’ kidney detection app. The ICO has required the Royal Free to give undertakings, including carrying out a privacy impact assessment (PIA) and commissioning a compliance audit. PIAs in particular are becoming generally used tools for data protection compliance management.
Although GDPR and data protection will be most in the spotlight in 2018, legal teams will need to continue to focus on the other elements of data law – data security (avoidance of and responding to data breaches), data sovereignty (control over data residency) and data permissioning (ownership and licensing) – to ensure appropriate management of data assets across the organisation.
In an October 2017 Review for the DCMS, ‘Growing the AI industry in the UK’, the authors’ first recommendation was that government and industry should deliver a programme to develop data trusts, as ‘a set of relationships underpinned by a repeatable framework compliant with parties’ obligations to share data in a fair, safe and equitable way’. In 2018, we’ll all be hearing more about data trust frameworks for managing data sharing and data governance. In general terms, the trust framework harnesses together for a particular data sharing use case (i) a set of overarching high-level ethical principles, (ii) operating rules on who can use the data and how, (iii) technical specifications and (iv) a legal framework on what happens if something goes wrong. Increasingly, these frameworks will combine legal terms – as contracts and notices – with functionality – as software embedded rules.
The trend towards standardisation of data governance at the international level also continues to gather pace. The International Standardisation Organisation (ISO) has done great work in 2017 in publishing as international supersets (i.e. to sit above but consistently with EU, US and other privacy rules) standards on data governance (ISO/IEC 38505-1) and PIAs (ISO/IEC 29134) and a code of practice for protecting PII (personally identifiable information – ISO-ese for personal data) (ISO/IEC 29151).
Growth of machine learning
Away from the headlines, expanding communications bandwidth (introduction of 5G, investment in subsea cables) and continuing 20% year on year growth in the Cloud provide the infrastructure for the digital transformation of the 4IR.
Incorporating an AI foundation in any major software development is increasingly taken as a given. In 2000, the internet ‘disappeared down a thousand foxholes’ into countless developments and applications before emerging ubiquitous as Web 2.0 and the [mobile+social+cloud] combination after 2004. Machine learning, as the 4IR’s ‘killer app enabler’ that can be trained to see patterns and make predictions, is at the ‘disappearing’ stage, with work going on around the world embedding it into existing applications and using it to develop new ones.
Machine learning costs have tended to be frontloaded with intensive, iterative use of training datasets, but new techniques like transfer learning (which pre-trains the model on similar data) and synthetic data (which uses computer generated data to mimic the real dataset) are set to significantly reduce lead times and on costs. At the processor level, specially developed AI chips (like Intel’s recently announced Nervana Neural Network Processor (NNP)) are also accelerating training times, and software developers are starting to pack machine learning models into mobile devices using greater compression and power efficiency.
As AI makes software cognitive and ‘learn to learn’, intelligent apps and analytics are starting to change the structure of work more deeply than previous technology waves; and intelligent things (whether static, like sensors: [machine learning+machine perception] or mobile, like autonomous vehicles and robots: [machine learning+machine perception+machine control]) are entering the ‘hockey stick’ phase of rapid growth.
The permeation of machine learning enabled software throughout business has a number of implications for technology lawyers. First, through techniques like DevOps (next generation Agile) software development is getting faster, with shorter cycles, higher deployment frequency, ‘containerisation’ of apps across environments and greater componentry re-use, so understanding the ownership and licensing arrangements of the whole code base to ensure correct permissioning is key. Second, through deployment techniques like SOA (service oriented architecture) and RPA (robotic process automation) software systems will increasingly communicate indirectly with many more other systems, so check the software licence grant clause to make sure calls on the licensed software by remote, indirect systems are not unexpected chargeable use. Third, check the ownership and permissioning position for the data that the software processes – whether training datasets, workload datasets or data derived from the software as analytics and insights.
Digital reality and Blockchain
In its annual tech trends report for 2018, ‘The Symphonic Enterprise’ Big4 accounting firm Deloitte highlights “digital reality, cognitive and blockchain [as] the stars of the enterprise technology realm [that] are redefining IT, business and society in general”.
The Deloitte report quotes a prediction from consultancy IDC that spending on digital reality – a mix of augmented reality (AR), virtual reality (VR), mixed reality (MR) and immersive technology – will grow from $9.1bn in 2017 to over $150bn by 2021 as we adapt to a work environment where, like our mobile today, our AR/VR/MR device is ‘always on’ to communicate, learn, train and interact. This shift, the report says, is comparable to that from “client-server to the web and web to mobile [leading] to more natural and intuitive ways for technology to better our lives. Indeed our means of interface with digital information will likely no longer be screens and hardware but gestures, emotions and gazes”. For lawyers, the digital reality era will bring to the fore copyright permissioning and rights legal issues about generating, communicating and adapting content. It will likely also add another dimension to copyright policy debate about Open Access and fair use.
In the Deloitte report, the blockchain – a distributed ledger that ‘trustlessly’ serves as a tamper-proof record of who owns what – has its place as a single component whose ‘unique’ is to record the exchange of assets in an open, secure financial transaction protocol and enable material improvements in security and accuracy across business processes like assets transfer registration, claims handling, trade finance and the supply chain. Smart contracts take the blockchain, as a protocol, to the next level, as a utility of general application.
As if each of these developments – machine learning, cognitive software, blockchain and digital reality – doesn’t represent change enough in its own right, it’s really in their combination – particularly how insights, analytics and automation from cognitive software can work with blockchain, smart contracts and digital reality – that truly radical transformation lies. But that’s probably one for 2019, not 2018.
 At page 4
 Ludwig Siegele, ‘The Dodd-Frank of Data’ (The World in 2018, the Economist)
 Speech to the Institute of Chartered Accountants in England and Wales, 17 January 2017 – https://ico.org.uk/about-the-ico/news-and-events/news-and-blogs/2017/01/gdpr-and-accountability/
 Royal Free – Google DeepMind trial failed to comply with data protection law – https://ico.org.uk/about-the-ico/news-and-events/news-and-blogs/2017/07/royal-free-google-deepmind-trial-failed-to-comply-with-data-protection-law/
 At page 2
 At page 75