Data Policy Digest

Gavin Freeguard

Gavin Freeguard

Hello, and welcome to our thirteenth (lucky for you) Data Policy Digest, bringing you all the latest data and AI policy developments. Of which there continue to be many.

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 gavin@connectedbydata.org. We’re on Twitter @ConnectedByData and @DataReform. 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, Digest #11 and Digest #12.

To receive the next edition of the Data Policy Digest direct to your inbox sign up here.

Data policy developments

Deeply DPDIB

Last time out, the rumours suggested the Bill would enter Lords committee stage in late February. We’ve now heard it will be late March - on Wednesday 20, Monday 25 and Wednesday 27 March, then Monday 15, Wednesday 17, Monday 22, Wednesday 24, and Monday 29 April (with an extra day on Wednesday 1 May if necessary).

Amendments are already being laid (a reminder that, in ‘grand committee’, they can only be agreed unanimously - otherwise they’ll be brought up during the debate on report stage). They include one on freeing the Postcode Address File and another on international data transfers and contractual clauses, which mover (and former ministerial shaker) Lord (James) Bethell has written about.

Regular readers will know we (and many others, including the Commons and the Lords) are concerned about the late and under-scrutinised addition of sweeping DWP powers to tackle fraud to the Bill. A petition on the new DPDIB proposals currently has just over 20,000 signatures - and has elicited a response from DWP (complaining about ‘misconceptions’, which could have been avoided had, say, the whole thing been scrutinised properly). While we’re on DWP, the Guardian reports that the department has told Jobcentres to stop referring benefit claimants to food banks, citing data privacy concerns. ‘Sceptical’ would be a kind word for some of the responses to that. ‘Disingenuous’ and ‘suspiciously conveniently timed’ would be others.

And… the Lords Delegated Powers and Regulatory Reform Committee - which reports on the appropriateness of powers government is giving itself - was not best pleased with some of the measures in the Bill… there was a carry-over motion for the Bill, which means it can continue to be debated until this parliament ends… and Baroness Kidron told the New Statesman she’d like to see the Bill ‘overhauled’.

We’ll be updating our resources page on the Bill as briefings, etc, emerge. And… the Open Rights Group have an online briefing THIS EVENING.

Bills, bills, bills

Online Safety Act Esther Ghey has called for further action to protect children, following the recent conviction and naming of her daughter’s killers… she did a big interview with Sky News (which they summarised)… Brianna Ghey’s mother and Molly Russell’s father join forces to combat online harm, reports the BBC… while Diginomica report on a recent conference touching on some of the issues - including some scepticism about whether some of the proposed measures would work as intended.

Also… the Crown Prosecution Service has convicted a first person under cyberflashing provisions (the BBC has more)… the ICO has issued guidance on content moderation and data protection… Politico looks at the ‘Great British Porn Block’ (not, as far as I know, a new Paul Hollywood/Prue Leith Channel 4 vehicle)… and ‘Ending hate on social media will cost billions’, Meta veteran Sheryl Sandberg has said.

Digital Markets, Competition and Consumers Bill It will be Report Stage next, with the government saying it will continue the conversation between now and then. Meanwhile the BBC reports that Microsoft denies it went back on its word on Activision job cuts (Digests 4 and 6 have more on where all of that came from).

Other bills Last time, we noted an amendment to the Victims and Prisoners Bill about a consistent victim identifier - Natalie Byrom has more on what happened next… the Investigatory Powers (Amendment) Bill has its Second Reading in the Commons today - the Commons Library has more… it’s Lords Third Reading for the Automated Vehicles Bill… and the Media Bill has its Second Reading in the Lords next week (the Lords Library has a briefing)…

Across the pond, ‘The Justice in Forensic Algorithms Act would allow defendants to access the source code of software used to analyze evidence in their criminal proceedings’… meanwhile, across the channel (well, channels - English and St George’s), the EU’s Digital Services Act came into force over the weekend - ‘This will see new responsibilities imposed on online platforms who have users in the EU, with the aim of better protecting those users and their rights’. TechCrunch, Politico and Amnesty are among those with takes.

AI got ‘rithm

Habemus charta alba! We have white smoke on the government’s AI white paper response, with some guidance for regulators on implementing the principles (Michelle Donelan has written to some regulators too), and a name change for CDEI for good measure. The press release and written statement to parliament have summaries.

We thought the (sensible) context specific approach is undermined by weak commitments, a lack of practical steps and support and lip service to public voice (Jeni also appeared on Channel 4 News); Ada thought it was ‘all eyes, no hands’, having warned ministers not to wait for a Post Office-style scandal before regulating AI; the Public Law Project welcomed the move towards mandatory public sector algorithmic transparency, something the ODI also cautiously welcomed while worrying about the lack of money and powers for regulators; some Cambridge experts say government should ‘aim policies at “hardware” to ensure AI safety’; while ‘Rishi Sunak dithers on AI as the US and EU forge ahead’ was Politico’s take. Tech academic Jennifer Cobbe’s exasperation may also strike a chord with many.

Over in the EU, the European Parliament voted on the AI Act last week. Politico has a cheat sheet on what it all means… in the US, the Biden-Harris Administration Announces First-Ever Consortium Dedicated to AI Safety (which includes civil society voices, praise be)… Japan’s ruling party pushes for AI legislation within 2024, Nikkei reports… the UN Secretary General says AI must not replace human agency… and we have a pope (habemus papem!) taking AI seriously according to recent profiles of his AI advisor.

We covered AI deepfakes in the context of Taylor Swift last time. There’s been much more discussion over the last fortnight, with tech giants coming together at the Munich Security Conference to make pledges around AI-generated deception… the US went into the conference worried that AI could ‘supercharge’ election disinformation, while Google’s Sundar Pichai said in advance that AI can strengthen cyber defences, not just break them down, and Politico’s preview of the conference noted ‘How tech-savvy civilians became the new masters of war’…

Politico’s Digital Bridge gave much of its edition last week to the subject, Deepfake porn is political violence… Carnegie Mellon’s Block Center for Technology and Society published a responsible voter’s guide to genAI and political campaigning… Sadiq Khan told the BBC that fake AI audio of him nearly led to serious disorder… Faculty’s Angie Ma told UKTN that AI is a double-edged sword in an election year… Facebook and Instagram are to label all fake AI images, reports the BBC… as Hugging Face publishes Provenance, Watermarking & Deepfake Detection: Technical tools for more control over non-consensual synthetic content… the Institute for Strategic Dialogue looked at ‘Spamouflage’ - pro-Chinese attempts to influence the US election… in the US, the FCC has banned robocalls with AI-generated voices… 404 Media used a police report to show how a high school ‘deepfake nightmare’ unfolded… Prospect argues that ‘Deepfakes are taking over reality’… and BBC News carries a story that Offenders [are] confused about ethics of AI child sex abuse…

VentureBeat reports that Sam Altman wants up to $7 trillion for AI chips. The natural resources required would be ‘mind boggling’… the FT reports that Data centres curbed as pressure grows on electricity grids (FT)… while Tech Policy Press argues that Measuring AI’s Environmental Impacts Requires Empirical Research and Standards…

The FT has taken a closer look at OpenAI’s finances - Can OpenAI create superintelligence before it runs out of cash? - while the company apparently has a new video generation tool out… Google Bard has become Gemini as they launch a new model… and another FT piece wonders if the AI hype has echoes of the telecoms boom and bust…

And… AI Companies Take Hit as Judge Says Artists Have “Public Interest” In Pursuing Lawsuits (Hollywood Reporter)… The text file that runs the internet (The Verge)… Researchers used artificial intelligence to decipher the text of 2,000-year-old charred papyrus scripts (Nature)… The messy landscape of artificial intelligence ethics and risk - a note for activists (Gail Bradbrook, picking up our Design Lab)…

The unsettling scourge of obituary spam (The Verge)… How an algorithm denied food to thousands of poor in India’s Telangana (Al Jazeera)… Parents turn to AI tablets after China’s tutor crackdown (Rest of World)… The software says my student cheated using AI. They say they’re innocent. Who do I believe? (The Guardian)… The Problem With Public-Private Partnerships in AI (Foreign Policy)… What the public think of charities using AI (CAF)… a new open access book, Algorithms of Resistance: The Everyday Fight against Platform Power…

And… Air Canada must pay damages after chatbot lies to grieving passenger about discount.

DSIT up and take notice

DSIT celebrated its first birthday with a cake… and reporting on progress on Science and Technology Framework. DSIT’s most senior civil servant talked to Civil Service World about setting up the department (and she’ll be talking to the Strand Group on ‘policy making for the age of AI’ in a few weeks).

While we’re on civil service digital leadership… the Central Digital and Data Office shakes up leadership as chief exec Lee Devlin to depart, reports CSW, while the Scottish Government seeks CTO to ‘set vision and direction for technology strategy.

But the big news is that, after 9 (NINE) years, the government has appointed a chief data officer - Craig Suckling, who joins from Amazon Web Services.

For those of you that haven’t taken an unhealthily close nerdy interest in the CDO saga… former head of GDS Mike Bracken was made CDO in 2015, but left government a few months later… the 2017 Government Transformation Strategy promised a CDO by 2020… government tried hiring a Chief Digital Information Officer in 2019… the 2020 National Data Strategy promised a chief data officer (though the details were silently changed shortly afterwards)… early 2021 brought a reconfiguration of government digital leadership and some initial confusion… some weirdo did a sea shanty about the whole thing… a promise a CDO would be appointed became much less committal a few months later… there was another attempt to recruit in 2022… and then again in 2023, which appears to have finally led to an appointment. Let’s hope it doesn’t take them quite as long to actually get anything done.

As for ministers… AI and IP minister Viscount Camrose has been in Brussels… new(ish) science minister Saqib Bhatti did a podcast about his experience of business and government… Chancellor Jeremy Hunt discussed AI with his German counterpart (finance minister, not Chancellor)… while Chancellor of the Duchy of Lancaster Oliver Dowden made a speech on cyber and commissioned an MP to review cybersecurity and economic growth…

Public Technology are doing a series of interviews in partnership with CDDO, including Treasury second perm sec Cat Little on how policy and digital can combine to deliver ‘radical’ new outcomes… and they’ve looked at how old government’s digital professionals are…

The head of the Office for Statistics Regulation has called for Whitehall data transparency to become ‘the norm’ in select committee evidence… as MPs warn that a delayed review of the UK Statistics Authority leaves it in an ‘unsatisfactory and unreasonable situation’… the UKSA is getting an advance on its funding… the ONS has blogged about how it is ‘transforming its statistics for local areas’… and given some more details about the datasets in its Integrated Data Service…

The government says data sharing powers on fraud and debt in the Digital Economy Act have saved taxpayers £137 million since introduction and is planning to retain them… as a report from the NAO found ‘Officials struggled to access data elsewhere in government to support Covid response’…

Faculty have published some learnings from ‘building generative AI-powered public services’… as their CEO argued the AI Safety Institute should focus on standards, not testing… and the AI Safety Institute published its approach to evaluations… DSIT published an introduction to AI assurance… the AI Opportunity Forum held its first meeting… ‘AI firms must address hallucinations before GOV.UK chatbot can roll out, digital chief claims’… as DWP bans officials from using ChatGPT…

South Korea is top of the OECD Digital Government Index… as part of their new AI Campus, Apolitical have interviewed Estonia’s chief data officer… Apolitical’s CEO thinks governments could learn some things from startups on AI… while some other digital government thinkers think the UK general election could help reset digital government…

There’s a consultation on health and social care statistical outputs (background)… the ICO is looking for an Expert Member of its Technology Advisory Panel (deadline: 23 February)… the Metropolitan Police is creating a head of data rights post to ‘ensure transparency and build trust’ (as the policing minister continues his facial recognition obsession)… DSIT is looking for a Head of AI Regulation Strategy & Implementation… Ireland has appointed two new data protection commissioners…

ICO published its latest Tech Horizons report… the Government Office for Science published a review of government science capability… and DSIT published minutes from the latest meeting of the Geospatial Commission board… AI Fairness Innovation Challenge winners announced… Introducing the Innovation Clusters Map… Research Ventures Catalyst: successful applications… government response to the Review of research bureaucracy… and - one of those things which sounds dry and technical but has long been a bugbear of digitally-minded civil servants and others across government - an independent report from Lord (David) Willetts on Reforming the DSIT business case process.

Parly-vous data?

Both Houses are back from recess - the Commons after a week off, the Lords after a couple of days away.

Let’s start with the Commons… data sharing is one of the main problems in trying to work across government, according to a new Public Accounts Committee report (chair, Meg Hillier, has also written about it).

Coming up… the Levelling Up committee is taking evidence on the data-heavy Office for Local Government… a subcommittee of the Defence select committee quizzes witnesses on ‘developing AI capacity and expertise in UK Defence’ on 20 and 27 February… the Transport select committee has an evidence session on the future of transport data on 21 February… there’s an adjournment debate on access to social media in prisons on 26 February… and DSIT questions take place this Wednesday 21 February (then again on Wednesday 17 April, with Cabinet Office questions next Thursday 29 February and then Thursday 25 April).

In the Lords… DSIT secretary Michelle Donelan and her DG for digital tech and telecoms gave evidence to the Communications and Digital Committee on the day the AI white paper response was published (though it was, ironically, beset by tech problems)… government chief scientific adviser, Professor Dame Angela McLean, was up in front of the Science and Tech committee a week later… the Lords’ Industry and Regulators Committee published ‘Who watches the watchdogs? Improving the performance, independence and accountability of UK regulators’, which touched on digital skills and overlaps…

On the floor of the House, there was a debate on digital exclusion… and discussions on questions from Lord Clement-Jones on computer misuse and computer evidence… Baroness Owen on AI ‘nudify’ apps… and Lord Strasburger on the National Security Online Information Team (formerly known as the Counter Disinformation Unit) - Public Technology have a write-up and Big Brother Watch a thread…

And there are evidence sessions coming (20 and 27 February, 5 March) on the Communications and Digital Committee’s future of news inquiry… and the Joint Committee on the National Security Strategy is launching a new inquiry into Defending Democracy, which of course includes some questions on AI - you have until 18 March (h/t Tom King).

Labour movement

Shadow DSIT secretary Peter Kyle was visiting tech companies in the US (the full list is at the end of the press release - the US does have some great civil society orgs and academic research centres too, you know).

Shadow science minister Chi Onwurah told OpenUK’s State of Open Conference that regulation can support rather than stifle innovation, and ‘A Labour government would urgently introduce binding regulation of a small group of companies developing the most powerful AI models’. (You can watch the whole conference, read a summary, and check out the other keynotes from Conservative Baroness (Tina) Stowell and Cambridge academic Neil Lawrence, too.)

There are also some titbits on policy priorities in this BBC article, Labour works on plan for power - not forgetting LabourList has a summary of the full National Policy Forum document from a few months ago.

And the first hustings took place in the Welsh Labour leadership contest.

On the Horizon

Horizon: Ministers to work with Fujitsu on compensation costs but has no plans for law to tell citizens of government IT errors (Public Technology)… Kemi Badenoch failed to raise Horizon scandal when she met Fujitsu at Davos (The Guardian)… and expect to see more ‘parallels with Horizon’ stories like this one, English test scandal: students renew fight to clear names after 10 years (The Guardian).

Oh and just when you thought things might have quietened down a bit… yesterday’s Sunday Times: Post Office boss: I was told to stall compensation to help Tories.

In brief

What we’ve been up to

What everyone else has been up to

Events

Good reads

And finally:

Data Protection and Digital Information Bill
 Previous

Need help?

Do you collect, use or share data?

We can help you build trust with your customers, clients or citizens

 Read more

Do you want data to be used in your community’s interests?

We can help you organise to ensure that data benefits your community

 Read more

Want to get involved?