Sign in

Llibertat UK (Katrina)

@llibertatuk.bsky.social
382 followers 928 following 289 posts

Design for humans. End to end, front to back, user-centred digital services with real impact for the public, private and third sectors globally. Led by former GOV.UK Head of Content Padma Gillen. Accessible by design. Posts by Katrina (she/her)

PostsRepliesMedia
Llibertat UK (Katrina) @llibertatuk.bsky.social · 07/10/2026
If you're leveraging AI to unlock new efficiencies but you've no idea what it does, why, or what the ROI is, you're just making word salad. Start with: - embedding AI safety first in your company culture - identifying and assessing risks - defining governance - testing, monitoring and iterating
010
Llibertat UK (Katrina) @llibertatuk.bsky.social · 05/10/2026
AI readiness isn't just about technology. It's about having the right infrastructure: - data - security - privacy - processes - accountability - people and skills Don't let the rapidly evolving capabilities of AI fracture your business. Don't scale what you don't understand.
000
Reposted by Llibertat UK (Katrina)
Tom Loosemore @tomski.com · 02/10/2026
My colleague Audree Fletcher has published a fabulous article in the Lewis Silkin magazine making the case for pragmatic optimism. Am big fan of Audree. And big fan of her ideas. www.lewissilkin.com/our-thinking...
lewissilkin.com
The hard-nosed case for optimism
<div class="ck-content"><p>Despite its reputation for softness, a particular kind of optimism is the most useful thing a leader can bring to bear in times of uncertainty.&nbsp;</p></div>
086
Reposted by Llibertat UK (Katrina)
Ada Lovelace Institute @adalovelaceinst.bsky.social · 01/10/2026
We're hiring! We are recruiting a Researcher to support our work on equity and inclusion in participatory and public attitudes research. www.adalovelaceinstitute.org/job/research... Closing: Monday 19th October 2026.
142
Llibertat UK (Katrina) @llibertatuk.bsky.social · 02/10/2026
What causes "AI misalignment". Not "rogue agents". Decisions by people. How to build AI safety: Human governance: permissions, approvals, testing, monitoring. Clean training data. Evaluations testing the right things. Deployment with controls. Company culture that centres not harming humans.
030
Reposted by Llibertat UK (Katrina)
Hannah Shelley, MLIS (Maggots, Lycanthropes & Ice Skeletons) @hannahshelley.site · 02/10/2026
Please stop anthropomorphizing computers. They cannot "run" a program, because they don't have legs
392066286
Reposted by Llibertat UK (Katrina)
Tom Loosemore @tomski.com · 17/09/2026
I made my first thing that's aimed primarily at AI Agents. Does anyone fancy helping me test it, by pointing their Ai Agent at it? If so, send me a DM (or just reply) and I'll ping you over details. Ideally, your agent will tread lightly, given it's very early days for my new thing.
Photo of an epaper fridge magnet showing what day your bins will be collected and what bins to leave out
262
Reposted by Llibertat UK (Katrina)
Konstantinos Komaitis PhD @kkomaitis.bsky.social · 02/10/2026
Part 4 is out: This part looks beyond individual controversies to the changing environment itself and asks whether the assumptions that once made platforms successful still fit the digital world they helped create. www.techdirt.com/2026/10/01/t...
techdirt.com
The Metric Is Not The Mission: The Digital Climate
The Metric Is Not the Mission is a ten-part examination of how Big Tech moved from building and expanding the open internet to increasingly shaping it around its own metrics, incentives and assumptions. Across the series, the argument follows the evolution of the platform economy—from the optimism of the early internet to the growing tensions...
1228
Llibertat UK (Katrina) @llibertatuk.bsky.social · 01/10/2026
Think your organisation is safe from AI dangers? Do you know: - How much content you have? - How much content is usable? - Who owns your content? - Who does what in a crisis? AI-readiness audits can help: - map your content estate - measure your risks - prioritise actions - report and monitor
000
Reposted by Llibertat UK (Katrina)
Accessibility Awareness @a11yawareness.bsky.social · 01/10/2026
Hyperlink text should make sense when read out of context. Screen reader users can navigate from link to link, and can listen to links in a list. When navigating this way, only the link is read. So "click here" or "read more" won't make sense.
04520
Reposted by Llibertat UK (Katrina)
Inclusive Design 24 (#id24) @inclusivedesign24.bsky.social · 01/10/2026
A week ago today, Inclusive Design 24 #id24 was in full swing. inclusivedesign24.org/2026/schedule/ Missed it live? No worries, all videos are available for free on our YouTube channel www.youtube.com/inclusivedes... #a11y #accessibility #inclusiveDesign
inclusivedesign24.org
Inclusive Design 24 (#id24)
A free 24-hour online community event on accessibility and inclusion
01311
Reposted by Llibertat UK (Katrina)
zeldman @zeldman.bsky.social · 01/10/2026
New HTML and CSS features you should be using in your design system per @zeroheight.com. zeroheight.com/blog/new-htm...
zeroheight.com
New HTML and CSS features you should be using in your design system - zeroheight
Declan Chidlow explores new HTML and CSS features that can make your design system more flexible, adaptable, and maintainable.
2519
Reposted by Llibertat UK (Katrina)
Ethan Mollick @emollick.bsky.social · 30/09/2026
As a heads-up: every firm's customer service agents are about to be overwhelmed with Dots & Muses negotiating for better deals using voice/chat channels made for humans. Stories of people delegating this sort of work to their agents & saving money are popping up and are going to only go more viral.
1017919
Reposted by Llibertat UK (Katrina)
Tech Policy Press @techpolicypress.bsky.social · 01/10/2026
The choice for Europe is not between regulating and building. It is between continuing to set rules for infrastructure that others operate, and acquiring the capacity to operate infrastructure whose rules Europe sets and can change, writes Nicholas Gates.
buff.ly
Europe’s Digital Sovereignty Agenda Must Include Digital Public Infrastructure
Europe has DPI capabilities, but has yet to organize them into a coherent strategy for digital sovereignty, writes Nicholas Gates.
044
Llibertat UK (Katrina) @llibertatuk.bsky.social · 29/09/2026
Intelligence isn't consciousness. AI can 'do' but not 'experience' things. Brains are not computers. "If we confuse ourselves too readily with our machine creations, we not only overestimate them, we also underestimate ourselves." Great long read by Anil Seth: www.noemamag.com/the-mytholog...
noemamag.com
001
Reposted by Llibertat UK (Katrina)
AI Tech News @aitechmatome-eng.bsky.social · 29/09/2026
The US government has required OpenAI and Anthropic to undergo domestic reviews before providing their advanced AI models to the UK's AI Safety Institute. US officials stated that ensuring the safety of domestic systems comes before sharing with allies. While the UK institute has previo...
記事のまとめ画像記事のまとめ画像
001
Reposted by Llibertat UK (Katrina)
Stacie Smith (She/Her) 🇨🇦 🏒 🧠 ⚾️ @stacesmith94.bsky.social · 29/09/2026
OpenAI abandons plan to release upcoming model as safety concerns escalate
cnbc.com
OpenAI abandons plan to release upcoming model as safety concerns escalate
The heads of OpenAI and rival Anthropic have both indicated recently that top AI labs should slow the pace of model development.
001
Llibertat UK (Katrina) @llibertatuk.bsky.social · 17/09/2026
Trust. That's the problem. It's easy for people to lose trust in technology, especially AI. So before you invest in AI, ask 3 key questions: 1/ What problem are you trying to solve with AI? 2/ What are the risks that AI will make it worse? 3/ How can you be AI ready without making risks worse?
000
Llibertat UK (Katrina) @llibertatuk.bsky.social · 10/09/2026
AI risk assessment isn't a one-off. Embed it across all the processes your people already use. Discovery, design, development, deployment, operations. That means governance all the way through - transparency, approval, monitoring. Risks don't start or stop - they're there all the way. Start now.
000
Reposted by Llibertat UK (Katrina)
Elle Cordova @ellecordova.bsky.social · 08/09/2026
I am the very model of a modern day typographer A font take on the tune “Modern Major General” by Gilbert & Sullivan #typography #fonts #graphicdesign
24550411565
Llibertat UK (Katrina) @llibertatuk.bsky.social · 09/09/2026
Are you AI ready as an organisation? What about your website? Or your content estate? How much content is there? Who owns what? Be honest. 80% of your content hasn't been touched in years, has it? AI readiness audits can help. Check what you own and what AI can use before you invest in any AI.
010
Reposted by Llibertat UK (Katrina)
Global Network Initiative @thegni.bsky.social · 08/09/2026
Building on our work from the India AI Impact Summit in February, the Working Groups are anchored in three critical pillars: 🔹 Safe & Trusted AI 🔹 Global AI Governance 🔹 Context-driven Infrastructure (2/6)
102
Reposted by Llibertat UK (Katrina)
Global Network Initiative @thegni.bsky.social · 08/09/2026
As the world prepares for the next round of global AI governance convenings - the UN Global Dialogue on AI Governance & the Geneva AI Summit - MAP-AI is excited to launch three Working Groups to develop concrete policy proposals feeding into these processes. 🧵 (1/6)
111
Reposted by Llibertat UK (Katrina)
nus-cil.bsky.social @nus-cil.bsky.social · 08/09/2026
How is the EU AI Act shaping AI governance beyond Europe? Join CIL’s Distinguished Speaker Series with Prof Christine Kaddous on 6 Oct, 2:00-3:45 pm at NUS Bukit Timah Campus. Register: cil.nus.edu.sg/event/artifi... #EUAIAct #AIGovernance #InternationalLaw
011
Llibertat UK (Katrina) @llibertatuk.bsky.social · 08/09/2026
AI governance isn't a policy. It controls product risks. Ask questions such as: - what could go wrong? - who's accountable? - what controls should exist? Build governance by providing answers through: - architecture - workflows - monitoring, testing, iterating Think governance by design.
000
Reposted by Llibertat UK (Katrina)
404 Media @404media.co · 07/09/2026
How does the U.S. government weaponize private data to surveil and kill abroad and at home? 404 Media is joining The Intercept for a LIVE discussion on this Thursday. Details below and join us here: join.theintercept.com/signup/Podca...
join.theintercept.com
The Intercept Briefing and 404 Media: A Live Podcast Event
Every week, The Intercept Briefing podcast covers the stories that the mainstream media can’t or won’t — from government spying to America’s forever wars. Join us for a special live taping featuring…
07223
Reposted by Llibertat UK (Katrina)
Stef Walter @stephaniewalter.bsky.social · 07/09/2026
My curated weekly newsletter is out. This week we cover accessibility ownership from design to development and why enterprise UX patterns matter for AI. Also, some tech ad archive, a dash of whimsical cat art, and a nine-tailed fox book. 👉🏻 stephaniewalter.design/blog/pixels-...
Pixels of the Week – September 6, 2026

Accessibility ownership, enterprise UX patterns for AI and whimsical cat illustrations.
111
Llibertat UK (Katrina) @llibertatuk.bsky.social · 07/09/2026
Don't risk AI damaging your business. Start by focusing on identifying data risks. Do you have historical data you can validate? Is your data structured? Is your data clean (accurate, consistent, complete)? Is your data scattered across systems?
000
Reposted by Llibertat UK (Katrina)
Tech Policy Press @techpolicypress.bsky.social · 01/09/2026
Regulators aren't treating AI as an exotic category outside privacy law, write Isabel Hahn and Aaron Alva. They're identifying where familiar principles get harder to operationalize as systems infer, remember, and act.
techpolicy.press
Privacy by Architecture Makes AI Compliance Work
Unlike the internet ad economy, where privacy and profit have often been in tension, AI presents an opportunity to align them, say Isabel Hahn & Aaron Alva.
2109
Llibertat UK (Katrina) @llibertatuk.bsky.social · 18/08/2026
Why spend money on being an AI-driven company if you don't get return on investment? Most businesses aren't seeing ROI from using AI. Why? Perhaps because, bluntly, it's AI for AI's sake. Start with why, what you need (including managing risk), and interrogate what value means.
000
Reposted by Llibertat UK (Katrina)
Accessibility Awareness @a11yawareness.bsky.social · 18/08/2026
You don't necessarily need to say "image of" in your alt text for users to know it's an image. Screen readers will announce that it's an image. But it can help readers to specify if it's a hand-drawn image, Polaroid, infographic, screenshot, chart, map, diagram, or so on.
09042
Reposted by Llibertat UK (Katrina)
Will Allen @willallennz.bsky.social · 08/08/2026
A narrow frame makes a complex issue look manageable. Agreed scope, defensible budget, signed-off plan. Most pressures push that way. But what gets left out keeps acting on the situation anyway. 🧪 New post on problem framing: learningforsustainability.net/post/problem...
learningforsustainability.net
Problem framing: why how we define issues shapes action
Problem framing shapes what becomes visible, who is involved, and what kinds of action seem possible in complex, multi-actor work.
0196
Reposted by Llibertat UK (Katrina)
Geoffrey Crofte @geoffreycrofte.bsky.social · 10/08/2026
After a viral comment on LinkedIn, the sudden attention motivated me to rewrite the blog post and clean up the dead links on this piece about accessibility. More specifically color (contrast). A kind reminder to designers that proper research is important. www.creativejuiz.fr/blog/en/user...
creativejuiz.fr
There is no “Myths of Color Contrast Accessibility” – Creative Juiz
Color contrast is not a detail you can skip. If you design interfaces, it is part of the job. And if you are used to low contrast ratios, yes, you are probably afraid of losing some of the beauty of t...
031
Reposted by Llibertat UK (Katrina)
Torrey Podmajersky @torreybird.bsky.social · 10/08/2026
I need a typeface that distinguishes lowercase L from uppercase I if you're going to post about Weird Al and AI at the same time 😂
031
Reposted by Llibertat UK (Katrina)
Stephan Faris @stephanfaris.bsky.social · 10/08/2026
Time Magazine has a separate version of its website with ads only AI can see www.theregister.com/ai-and-ml/20...
theregister.com
Time Magazine has a separate version of its website with ads only AI can see
Brands are already paying to influence what chatbots say about them
45638
Reposted by Llibertat UK (Katrina)
zeldman @zeldman.bsky.social · 10/08/2026
“All em-dashes in these slides were human-generated.”
Photo of a man at a podium, standing in front of a giant screen containing a presentation he is making to an off-camera audience. At the bottom of the screen, a footnote says "All em-dashes in these slides were human generated." A headline reads "Em dash disclaimers are now showing up in presentations" (essentially narrating what the photo already shows us).
411210
Llibertat UK (Katrina) @llibertatuk.bsky.social · 10/08/2026
Organisations are racing to adopt AI, but lack of infrastructure means accessibility gets left behind. Asking these 3 questions can help: - who owns accessibility in AI projects? - have you got solid data about users with different needs? - how will you develop rigorous formal governance?
000
Reposted by Llibertat UK (Katrina)
Accessibility Awareness @a11yawareness.bsky.social · 29/07/2026
Overlays are widgets or technologies that promise to improve a site's accessibility with third-party source code. More than 1,000 accessibility advocates and web developers have signed an open letter to explain the limitations and dangers of these products. overlayfactsheet.com
overlayfactsheet.com
https://overlayfactsheet.com/en/
0118
Reposted by Llibertat UK (Katrina)
illumi @yallchemist.bsky.social · 24/07/2026
If you’ve joined Bluesky in the past year or so you may not know this but a Bluesky tweet is called a skeet. Thank you for coming to my Ted Talk.
51798148
Reposted by Llibertat UK (Katrina)
Will Allen @willallennz.bsky.social · 29/07/2026
AI ethics debates often sound polarised because people talk at different levels. I find it useful to distinguish three: the conditions shaping AI, organisations introducing it, & everyday practices where judgement, participation and responsibility sit.🧪 learningforsustainability.net/post/ethical...
learningforsustainability.net
Ethics around AI use
An exploration of the wider ethical picture of AI, acknowledging both structural conditions and practice-level tensions.
1104
Reposted by Llibertat UK (Katrina)
Nexer_Digital @nexerdigital.com · 29/07/2026
🛍️ 87% of disabled shoppers told us they can't complete a typical retail journey without encountering accessibility barriers. #TheHiddenJourney report explores disabled people's experiences of shopping, and how organisations can improve experiences for everyone. www.nexerdigital.com/the-hidden-j...
A blurred photograph of someone using a credit card beside a laptop is overlaid with text highlighting findings from The Hidden Journey research. The text reads: "81% of disabled shoppers told us they experience accessibility barriers online, while 87% said they cannot complete a typical retail journey without difficulties."
001
Llibertat UK (Katrina) @llibertatuk.bsky.social · 29/07/2026
Accessibility debt is bad enough. Racing to adopt AI also creates AI accessibility debt, including: - embedded bias from training data - rapid reproducing of inaccessible components and patterns - trust and control issues, for example, with AI-assisted actions
110
Llibertat UK (Katrina) @llibertatuk.bsky.social · 23/07/2026
It's not about making your content 'AI ready'. You need to build organisational resilience so your content can work with any technology (because technology will keep changing). Develop strategic content infrastructure - governed, structured, modelled, owned, and readable by machines and humans.
020
Reposted by Llibertat UK (Katrina)
Institute for the Future of Work @ifow.org · 16/07/2026
In particular, Oliver highlighted what our research shows about the changing skills landscape - from growing numbers of skills required within roles to the increase in employer demand for human-centric skills.
ifow.org
Old Skills, New Skills - what is changing in the UK labour market? - IFOW
How is demand for skills changing in the UK? Using data from 2016 and 2022, this report reveals which sectors are experiencing most flux.
011
Reposted by Llibertat UK (Katrina)
Greg Dunlap @heyrocker.com · 16/07/2026
@koziarz.com www.phoronix.com/news/Microso...
phoronix.com
Microsoft Goes Nostalgic In Newest Open-Source Drop: Comic Chat Open-Sourced After 30 Years
Introduced in 1996 with Internet Explorer 3.0, Microsoft Comic Chat provided comic-like avatars driven IRC chat client
121
Reposted by Llibertat UK (Katrina)
Frankie Roberto @frankieroberto.com · 16/07/2026
The GOVUK design system team have today added the interruption page pattern and component: design-system.service.gov.uk/patterns/int... - based on previous work by teams across government. 🎉 This is a great example of where it is sometimes helpful to add some friction into a digital journey.
design-system.service.gov.uk
Interruption pages
Pause the user journey to give them important information
0186
Reposted by Llibertat UK (Katrina)
Accessibility Awareness @a11yawareness.bsky.social · 20/07/2026
Plain language is communication that your audience or readers can understand the first time they hear or read it. The terminology you use will vary depending on the audience. Always aim to avoid jargon when possible, especially when you have a broad general audience.
0147
Llibertat UK (Katrina) @llibertatuk.bsky.social · 20/07/2026
Purpose, value, outcome - identifying what people really need first. Not yet more stuff (that's most likely making already shaky governance even shakier).
000
Reposted by Llibertat UK (Katrina)
Hetan Shah @hetanshah.bsky.social · 19/07/2026
Like this from @timharford.ft.com on how inclusive design principles can often help all users, not those just seen as traditionally excluded www.ft.com/content/4d8e...
Subtitles help people who can't hear what the actors are saying - and it turns out that, for some gritty dramas, that is all of us. Web pages that are designed with screen readers for the blind in mind are typically better structured, simpler and friendlier for mobile devices. The Oxo kitchen appliance company was founded by a retired cookware entrepreneur whose wife had arthritis, but the chunky rubber handles are far more widely enjoyed. And of course, the dropped kerb may be life-changing for wheelchair users but it is pretty convenient for anyone pushing a pram.
The challenges faced by users who are unusually small, large, old or young, or who have disabilities, are, writes Epstein, "more extreme versions of the challenges that
many other users face"
1024878
Llibertat UK (Katrina) @llibertatuk.bsky.social · 20/07/2026
AI doesn't cover the cracks in a large, poorly governed website. AI makes the cracks obvious. Website chaos is more chaotic, front and back, from confused ownership to content no one needs. Strategic content infrastructure is essential - not just for AI outputs, but productive outcomes at scale.
110