Hello, and welcome to our twelfth Data Policy Digest, bringing you all the latest data and AI policy developments. Rest assured, this is one DPD that won’t use AI to swear at you. (Though may have you swearing at AI.)
If there’s something we’ve missed, something you’re up to that you’d like us to include next time or you have any thoughts on how useful the Digest is or could be, please get in touch via firstname.lastname@example.org. We’re on Twitter @ConnectedByData and Mastodon @ConnectedByData@social.coop. You can also catch up on Digest #1, Digest #2, Digest #3, Digest #4, Digest #5, Digest #6, Digest #7, Digest #8, Digest #9, Digest #10 and Digest #11.
Data policy developments
Like a lovelorn valentine, we’re looking for a date. Specifically, when the Bill will start Grand Committee stage in the Lords (more on what that means in our last edition). We’ve heard rumours of 21 or 26 February.
In the meantime, get yourself up to speed with our summary of Lords Second Reading - the first opportunity for peers to debate the Bill (tl;dr: they’re not happy with a lot of it and think it needs a lot of work). The whole debate is in Hansard for the real nerds (which, let’s face it, you are one of if you’re reading this).
A lack of dates does not mean a lack of activity… Lords have started laying amendments… the Lords Constitution Committee has published a report with concerns at far-reaching powers, lack of scrutiny and lack of explanation… the government has already backtracked on its backtracking on its promise to bereaved families around coroner access to data (BBC: Bereaved parents win online harm battle)… interesting to see briefing around the police time ‘wasted’ owing to ‘EU rules’ in the Mail… digital service assessments from 2022 have just been published for the National Underground Asset Register, recently added to the Bill (it failed, then met, its alpha stage)… How much extra spam will the UK’s Data Protection and Digital Information Bill create? asks Peter Wells…
Tens of thousands of people have signed Big Brother Watch’s petition against DWP bank snooping powers, says the Big Issue… those powers ‘would treat disabled people like criminals and further destroy trust’, campaigners told the Disability News Service… and a reminder that another controversial set of amendments, on ‘democratic engagement’, fits into a broader set of worrying electoral rule changes…
And… the former Biometrics and Surveillance Camera Commissioner once again took the opportunity to criticise the Bill (and government’s lack of engagement, and the Home Office), as a missed opportunity and the abolition of his (former) office a mistake. He also reveals the original expectation was that the Bill receiving royal assent in March 2023.
Bills, bills, bills
Online Safety Act The BBC’s ‘Sunday with Laura Kuenssberg’ had a strong focus on online safety_ _with the mother of murdered teenager Brianna Ghey and other guests including education secretary Gillian Keegan, and shadow DSIT secretary, Peter Kyle… Politico has a short summary… Banning children from social media would be short-sighted, Baroness Kidron wrote last week for PoliticsHome…
Parts of the Online Safety Act have come into effect - ‘Cyberflashing, epilepsy-trolling and fake news to put online abusers behind bars’, notes DSIT… a list of overseas regulators has been laid in the Lords… Meta whistleblower Arturo Béjar has been in London, saying regulators are the ‘last hope’ for fixing social media (a Guardian leader also picked up on his visit)… while Mark Zuckerberg and others were in front of Congress.
Digital Markets, Competition and Consumers Bill The Bill is making its way through Lords committee, with a couple more hearings this week… their Lordships are not happy about Secretary of State powers here, either.
Investigatory Powers (Amendment) Bill We await a date for Commons Second Reading (this one has already been through the Lords)… Big Brother Watch, ORG, Rights and Security International, Liberty, Internet Society, and Privacy International have published a critical joint briefing… while techUK and Apple are among others with concerns.
Other bills Lord Bach has laid an amendment to the Victims and Prisoners Bill about a consistent victim identifier.
AI got ‘rithm
More looking for dates - in this case, the much-delayed government response to its AI White Paper. Politico thinks it could be Tuesday - i.e. TOMORROW - the same day DSIT secretary Michelle Donelan is in front of a Lords select committee on generative AI and digital exclusion. Politico and the FT hear that plans for a voluntary code on the issue of copyright and AI, which the Intellectual Property Office had been leading on, have been dropped; that ‘targeted, binding measures will be needed for the most advanced “highly-capable general purpose models” (aka foundation models)… eventually’; and to look out for ‘a new steering committee to monitor and formalize coordination across regulators, updates on DSIT’s “central function”’ and consultations on a risk register and other parts of the regulatory system.
Over in the EU… political agreement being reached on the AI Act in December didn’t stop controversy rumbling on through January. (The Ada Lovelace Institute urged approval and EDRI published some useful background.) Member states reached agreement on 2 February. The Commission has also issued a decision establishing the European AI Office.
Lots of discussion of democracy, deepfakes and data… there’s a new book out on data-driven campaigning… the FT reported that ‘Audio deepfakes emerge as weapon of choice in election disinformation’… Tortoise published some views on how to avoid an AI disaster at the next UK election… and MPs ‘call for action on deepfakes as fears grow among MPs over election threat’… and…
More deepfakes (Taylor’s version). AI-generated explicit deepfakes of Taylor Swift brought the issue to the fore… NBC and the New York Times have summaries of the story which led to X/Twitter blocking the images… as does a Twitter account that pretended to be a cybersecurity-interested Taylor Swift, because we live in the future, or something… ‘It’s men telling a powerful woman to get back in her box’, notes the Guardian in their exploration of the scandal… Platformer see the affair as a warning, while Control AI polling found a lot of support for banning deepfakes… in general, outlets questioned the Blank Space where regulation should be and wondered how deepfake victims without a large fandom and international profile would Shake It Off… as an Australian politician also found herself at the centre of a doctored image scandal…
‘Are you ready for Elon Musk to read your mind?’, asked the New Statesman, coincidentally just before Musk announced a trial of the Neuralink brain chip… Nature asked scientists what they thought of it, while the New Scientist has an explainer… The Metaverse Flopped, So Mark Zuckerberg Is Pivoting to Empty AI Hype (tell us what you really think, Rolling Stone)… The Economist marks Facebook’s 20th birthday by announcing ‘The end of the social network’, while the BBC looks at four ways it changed the world… the US FTC has launched an inquiry into AI deals by Microsoft, OpenAI, Google and Amazon… while Gary Marcus looks at OpenAI’s 2024 challenges… ‘Google Splits Up a Key AI Ethics Watchdog’, its responsible AI team… and Anthropic worries that ‘AI poisoning could turn open models into destructive “sleeper agents”’…
AI is producing ‘fake’ Indigenous art trained on real artists’ work without permission (Crikey - which is the name of the news outlet as well as an understandable response)… while Damien Hirst and Tracey Emin among thousands of British artists used to train AI software, Midjourney… and ‘A New Nonprofit Is Seeking to Solve the AI Copyright Problem’…
There are plenty of jobs going at the UK’s AI Safety Institute, while some of those already attached to it have published a paper on how ‘Black-Box Access is Insufficient for Rigorous AI Audits’… the Parliamentary Office of Science and Technology have published a new POSTnote on AI in education… the World Economic Forum’s AI Governance Alliance has a new series of briefing papers… Ethan Zuckerman writes about the ‘Two warring visions of AI’ for Prospect… AI Now asks ‘What can we learn from the FDA model for AI regulation?’… and… ‘The tyranny of the algorithm: why every coffee shop looks the same’ - the Guardian runs a book extract on how algorithms are flattening culture.
DSIT up and take notice
Happy first birthday to the Department for Science, Innovation and Technology, established on 7 February 2023.
Over the past few weeks, the department has announced that the ‘State of the Science’ report (revealed at the AI Safety Summit) will be renamed the International Scientific Report on Advanced AI Safety, and unveiled an international advisory board for it (first edition expected this spring)… Politico also had the scoop that Google DeepMind’s Geoffrey Irving and University of Oxford academic Chris Summerfield have been appointed as research directors - turns out that was a sneak preview of the now-published third progress report from the AI Safety Institute… Michelle Donelan has been in Canada agreeing UK-Canada tech collaboration (and asking generative AI to suggest British snacks to gift)… we have some more details on the new AI Opportunity Forum… and there’s some more investment in quantum…
There’s been plenty going on with data and AI in government. Having trailed it way back in December, I managed to miss the actual publication of the UK government’s new generative AI framework, which coincided with our last Digest… Civil Service World have a summary… while GDS have published blogposts on ‘The findings of our first generative AI experiment: GOV.UK Chat’ and ‘Experimenting with how generative AI could help GOV.UK users’… and while we’re on AI, ‘Universal Credit claims no longer paused while AI fraud checks carried out’…
Minister for the Cabinet Office, John Glen, touched on AI and other digital reforms in a speech at the Institute for Government… the IfG’s latest Whitehall Monitor has a handy summary of the past year in government data and digital… while Reform and Deloitte’s State of the State asked public sector leaders what they thought (some AI scepticism, some digital progress but still real problems around data)…
CDDO CEO (chief exec of the Central Digital and Data Office, Megan Lee Devlin, for those of you not fluent in abbreviation) is departing and a new exec director is being recruited (we’ve still not heard whether a new chief data officer has been appointed, as far as I’m aware)… CDDO published a new Digital Readiness Check, a set of questions ‘for use by policymakers and others working in multidisciplinary teams on new policies or initiatives in government’ (written up by Public Technology)…
The longlist for the Civil Service Data Challenge has been announced… as nominations open for the Royal Statistical Society’s 2024 award for statistical excellence in trustworthiness, quality and value… as Cabinet Office celebrates the number of civil servants taking part in the ‘One Big Thing’ data learning initiative (Civil Service World have a write-up, too)… the chair of the Public Accounts Committee wants clear digital goals for new permanent secretaries…
Education secretary Gillian Keegan set out the ‘importance of evidence and innovation in approach to AI’ at Bett (British Educational Training and Technology Show) 2024… the Ministry of Justice have blogged about the progress they’re making against their data strategy… the National Cyber Security Centre have a report on The near-term impact of AI on the cyber threat, while their old boss Ciaran Martin has written about the British Library hack… the independent Comisiwn on the future of Wales noted how a lack of data is ‘a significant barrier to accountability and performance improvement’ and ‘there should be a presumption in favour of compiling, sharing and publicising service data as a matter of routine’… Public Technology reports that digital is ‘becoming much more of a political issue’ for local government…
And there’s an open consultation on ‘Protecting and enhancing the security and resilience of UK data infrastructure’, and one on alcohol licensing and age verification which includes the use of emerging technologies.
As mentioned above, DSIT secretary Michelle Donelan (and Emran Mian, Director General for Digital Technologies and Telecoms) will appear in front of the Lords Communications and Digital Committee on generative AI and digital exclusion.
The committee has been keeping very busy… ‘UK will miss AI goldrush unless Government adopts a more positive vision’ was the main headline from their new report into LLMs. ‘The Committee considered the risks around LLMs and says the apocalyptic concerns about threats to human existence are exaggerated and must not distract policy makers from responding to more immediate issues’… they’re also accepting evidence on ‘The future of news: impartiality, trust and technology’ (again with a heavy AI element) until 12 February… on 8 February there’ll be a Lords debate on their report on digital exclusion… and they’ve had some changes to their membership, as did the Science and Technology committee…
Sticking with the Lords, the Justice and Home Affairs Committee has written to the Home Secretary questioning the legality of facial recognition technology… there was a question on innovation and regulation… and there are forthcoming questions on what steps the government is taking to protect freedom of expression in the course of their work on combating disinformation (Lord Strasburger, Lib Dem - something Big Brother Watch have done plenty of work on)… and what action they are taking to reform the Computer Misuse Act 1990 to enable legitimate independent testing of computer systems (Lord Clement-Jones, Lib Dem).
It’s quieter in the Commons, though the Culture, Media and Sport select committee touched on the impacts of AI as part of their inquiry into ‘the current state of the UK’s Film and High-End TV industries’… there was an adjournment debate on AI scams and consumer protection, and a Westminster Hall debate on real time bus information in the north east… and there will be question times in the House for DSIT ministers on Wednesday 21 February (and Wednesday 17 April), and for Cabinet Office ministers on Thursday 29 February (and Thursday 25 April).
Lots of news about Labour’s manifesto as an apparent drafting deadline of 8 February hives into view. ‘Labour ditches radical reforms as it prepares ‘bombproof’ election manifesto’, is The Observer’s take. There’s no explicit mention there of tech policy, but the new deal for working people mentioned does include a right to switch off and (from 2022) some more detailed thoughts on collective bargaining and the use of tech in the workplace. Manifesto plans will be vetted by various key figures, including Jonathan Ashworth (shadow Cabinet Office) to see how proposals will withstand attacks, and Lucy Powell (shadow leader of the Commons, and former shadow DCMS) to see if they’ll make coherent legislation.
Technology got a shout out in Keir Starmer’s big speech to business leaders. Labour’s plan for financial services also talks about setting international standards for the use of AI, as part of a wider Labour AI strategy. There’s also a new open letter (is that the furthest we’ve got into a Data Policy Digest before mentioning one?) and grouping from Technology Entrepreneurs, Investors and Executives Supporting The Labour Party.
There was also some praise for Labour’s tech approach in the New Statesman from… recently departed (Conservative) science minister. George Freeman. He also wrote about the importance of a cross-party consensus on science, innovation and tech on his new Substack (in a post likely to be remembered most for its mortgage revelation).
Both Welsh leadership contenders have published manifestos. Jeremy Miles’ includes better use of data and AI in health (including prevention, and scrutinising where services are not meeting the needs of minorities) and building ‘a dedicated data science and Artificial Intelligence capability within the government’, while Vaughan Gething’s includes innovation in healthcare and a digital pass for public services (and possibly more, but most of the PDF isn’t searchable)… and…
Keir Starmer’s speech to the Civil Society Summit talked about the importance of partnership and empowering communities… Peter Kyle is on a visit to the US this week ‘to discuss safely exploiting tech innovation in a fast changing world’, from where he spoke to Laura Kuenssberg (see Bills, bills, bills, above)… there’s some more digital, data and tech detail (on top of what was already in the party’s health mission) in a new Child Health Action Plan… Peter Kyle was doing the rounds talking about Labour’s life sciences plan… and Ofcom chief Melanie Dawes and former DCMS perm sec Sarah Healey are among those profiled in the FT as possible Labour choices for the UK’s next top civil servant.
On the Horizon
Our very own Jeni asked What does the Post Office scandal teach us about data and AI regulation? (quoted by Politico in their UK Morning Technology newsletter)… ‘Time for data protection law reforms to honour the postmasters’, wrote Natalie Byrom in The Times… The Big Issue looked at how the scandal proves why workers’ needs must be put ahead of AI and technology… Nature ran a piece entitled ‘Computers make mistakes and AI will make things worse — the law must recognize that’… and as noted above, Tim Clement-Jones has a Lords question on independent testing of computer systems which feels rather Horizon-flavoured.
- Amazon fined for ‘excessive’ surveillance of workers (BBC News)
- Public back ‘patient passports’ to share medical records with any doctor (The Times - front page splash if you can’t mount the paywall)
- ‘Rapid innovation offers the promise of a new era of prosperity, but instead risks exacerbating trust issues, leading to further societal instability and political polarization’ - that’s the verdict of the new Edelman Trust Barometer, subtitled ‘innovation in peril’. (Maybe now government will believe us when we say you need to build trust in order to innovate… a good job they’re not doing anything at the moment to threaten that)
- EU states accuse TfL of huge data breach over clean air zone penalties, with many given to compliant vehicles (The Guardian)
- Fortnite to come back to iPhones in EU as Apple must allow its customers to access alternative app stores on iOS devices purchased in the EU from March (BBC News)
- Donald Trump’s ‘sex and bribes’ data protection case rejected by UK court (The Guardian)
What we’ve been up to
- A reminder to sign up to our virtual workshop on 13 February, ‘Datafied’: How data and AI policy will impact all areas of progressive action, and what civil society can do about it. Spread the word!
- We were pleased to support the research and writing of the Wales’ TUC report, ‘A snapshot of workers in Wales’ understanding and experience of AI’. Jack Sargeant MS, who hosted a launch event, has also written about the topic.
- And we held a Connected Conversation on the re-use of health data, following our report with Just Treatment, ‘Our health data stories’
What everyone else has been up to
- Demos published a new report on Generating Democracy: AI and the coming revolution in political communications
- Digital Action’s newsletter looks at elections and big tech - as if you needed another data-related newsletter in your life (there are plenty more in my list of data newsletters, podcasts etc if you do)
- The ODI wrote about AI everywhere all at once for Computer Weekly, and published pieces on data discovery with LLMs and Privacy Enhancing Technologies
- Ada have launched a new blog series on public participation and policy, starting with meaningful public participation and AI
- The IfG looked at the benefits of greater transparency to government (summary) - and right on cue, there was a Lords question on Freedom of Information last week
- Projects by IF have come up with some design patterns AI designed to empower people
- techUK are looking for a Head of AI, Data Analytics and Digital ID (10 months Fixed-Term Contract Maternity leave cover)
- Reset have had a reset
- Careful Trouble have a new report on Tech for Today - and for Tomorrow: building secure foundations for digital Britain, and have also looked at how ‘the market is failing – internet access must be affordable for everyone’
- A new report from the Tony Blair Institute looked at health/genomics data
- OpenUK have a new report on AI and Open Innovation
- There’s a new book on ‘The Ethics of Privacy and Surveillance’ by Carissa Véliz
- And Significance, the official magazine of the Royal Statistical Society, is serialising something I wrote about where the Covid-19 ‘R number’ - remember that? - came from (this also gives you a flavour of what to expect).
- Another plug for the IfG’s 48th Data Bites, a smart data special, on 26 February
- Open data day is approaching
- DSIT perm sec, Sarah Munby, lectures on ‘Policy Making for the Age of AI’ at the Strand Group on 6 March (you can also catch up on her first lecture, about DSIT’s potential)
- The Centre for Intellectual Property and Information Law (CIPIL) will hold its Annual Spring Conference entitled ‘Data Protection Reform’, in March
- Geoffrey Hinton is one of the speakers at ‘The politics and philosophy of AI’ alongside Dr Kate Vredenburgh, LSE, 19 March
- The Turing’s big AI UK conference takes place 19-20 March
- Unlock Democracy have a virtual event on Democracy and AI tomorrow
- The data collection app at the heart of the BJP’s Indian election campaign (Rest of World)
- Do disappearing data repositories pose a threat to open science and the scholarly record? (LSE)
- UK police facial recognition explained: What you need to know (Computer Weekly)
- Resisting Dehumanization in the Age of ‘AI’ (Emily M. Bender)
- Of top-notch algorithms and zoned-out humans (Tim Harford, FT)
- ‘Mad’ AI risks destroying the Information Age (Telegraph)
- My McLuhan lecture on enshittification (Cory Doctorow)
- THE RISE OF TECHNO-AUTHORITARIANISM (The Atlantic)
- Are we entering a “Data Winter”? (Stefaan G. Verhulst)