The latest issue of Pulse is now live. Here’s what shaped this month’s edition.
July was a month of visible motion in AI governance. New taskforces were stood up, departments reshuffled, guidance issued, and pledges signed on both sides of the Atlantic. Yet much of the most instructive reading in this month’s Pulse points in a quieter direction — towards the distance between the rules being written and the practice on the ground.
An implementation gap comes into view
Two assessments set the tone. The second edition of the Global Index on Responsible AI, covering 135 countries, finds responsible-AI governance expanding but unevenly: frameworks show evidence of implementation in only around half of cases, and fewer than one in five governments disclose their own use of AI. The UN’s Independent International Scientific Panel on AI, in its first Preliminary Report, reaches a complementary conclusion — that current safeguards are not keeping pace with the technology’s advancing capabilities.
Read together, they describe an implementation gap that no volume of new legislation closes on its own. It is a useful lens for the rest of the edition.
Motion at the centre of government
The regulatory activity in July was real, if uneven. In the EU, the AI Omnibus entered into force with targeted simplifications to the AI Act — extended timelines for high-risk systems, wider regulatory sandboxes, lighter administrative burdens — while the Commission issued guidance ahead of transparency obligations that begin to apply from 2 August.
In the UK, a new PM’s AI Taskforce took shape at the centre of government even as departments were restructured and DSIT was closed, its responsibilities for science, innovation and technology redistributed. A voluntary cyber resilience pledge, signed by more than sixty organisations, and fresh calls for evidence on data regulation rounded out a busy month. The pattern is familiar: institutions moving quickly, with the harder question of durable implementation left open.
Sovereignty, reconsidered
Questions of sovereignty ran throughout — in Stanford HAI’s survey of commercial sovereignty offerings, the EU AI Office’s competitiveness findings, and Canada’s updated digital sovereignty standard. The Stanford work is a helpful corrective: most sovereignty products reconfigure rather than eliminate dependence on a handful of upstream providers, which reframes the challenge as calibrating interdependence rather than escaping it. Several articles in the edition test the same idea against middle powers, from Türkiye to the wider platform economy.
Where Security and AI Converge
The convergence of cybersecurity and AI was another clear thread — from the EU’s Action Plan on Cybersecurity and AI to the White House’s “GOLD EAGLE“ clearinghouse for coordinating vulnerability detection across critical infrastructure. The Government of Alberta case study, in which Claude Code was used to scan hundreds of millions of lines of government code, offers a concrete example of the same shift moving from principle into practice.
The Human Side of AI
Alongside the institutional story, this edition carries a growing body of research on the closer, more personal dimensions of AI — companionship, creativity, and the uncertainties of human–AI relationships. New studies examine the fragility of AI companionship, the “authenticity gap” between human and generative creativity, and the technostress that can form around increasingly autonomous agents. Together they are a reminder that governance is not only a matter of states and standards, but of how people live alongside these systems day to day.
The foundations underneath
If there is a common thread, it is that frameworks alone do not govern — conditions do. That is the starting point of the new DSA Working Paper, Governing Well in the Algorithmic Age: The Foundations of Digital Statecraft, which introduces digital statecraft as the organising concept for an emerging field and sets out ten principles, each naming a condition whose absence produces a structural governance failure. It is the intellectual framework underpinning the DSA Cambridge Fellowship, whose 2027 cohort is now open for applications.
The same questions will be the focus of Data for Policy 2026 in Barcelona (8–10 September), convening researchers, policymakers and innovators around “Governance of/with AI: Implications for Data, Infrastructure, and Tech Sovereignty.” Registration is open until 25 August.
Read the full July edition of Pulse for the complete selection of reports, articles, blog pieces and community news.
Stay informed on the latest developments and insights in data and AI policy, subscribe to our newsletter here.