About
Freelance journalist and editor, based in Europe. Currently Euractiv's Tech Editor, after…
Services
Activity
6K followers
-
Natasha Lomas shared this"This is also what decentering AI means in practice: refusing the premise that AI is an exceptional technology around which all other politics — labor, sustainability, democracy — must bend. Instead, it is a technology that can, and must, be shaped around the constraints of what kind of societies we want to have." -- clear thinking on AI's cutting edge from Frederike Kaltheuner Leevi Saari https://lnkd.in/eYXX_3YnNatasha Lomas shared thisA Transformative AI Strategy for Europe is making the rounds today and here's what you need to know: This isn't just a strategy on AI - it's a full policy agenda on social protection, economics, and institutions. But the whole piece rests on one assumption: sovereignty, security, and prosperity hinges on access to closed models only a handful of US firms can build. If you accept that assumption, then Europe's options shrink to various degrees of dependency that's far worse than anything we know today. What you need to know is that serious experts fundamentally disagree whether the entire strategy's premises are even plausible. Serious policy making simply cannot reorganize society around a single bet that only holds if the very same US companies we're being warned about keep their pinky promise. The Financial Times, for instance, published an entire analysis last week that asked how big a threat open models are to AI hyperscalers (link below). And if you want to know what Europe can do instead, Leevi Saari and I have some clear recommendations, none of which require owning every layer of the stack (also link below).
-
Natasha Lomas reposted thisNatasha Lomas reposted thisOn 8 September 2026, Jacob Coxon resigned from Anthropic, warning that AI could end humanity before the decade is out. Evan Hubinger, another Anthropic researcher, put his estimate above 1 in 10 over the next ten years. The WSJ reported Coxon’s departure without independently assessing the extinction claim. Four days later, Amodei, backed by Altman and Musk, called for a slowdown in frontier AI development. The Economist has repeatedly carried his warnings. Old news. Musk signed a pause letter in March 2023, but the requested pause never occurred. Altman, Amodei and Hassabis signed an extinction warning that May, then kept building, with gusto. More warnings have repeatedly accompanied the race. One counter-signal ran the other way. In July 2023, more than 1,370 signatories backed a British Computer Society letter arguing that AI is not an existential threat and calling for standards and regulation. I signed it. I might have missed it, but I have found no coverage in either the WSJ or The Economist. Quousque tandem? How long will companies and the media warn about an AI-pocalypse while regulation is opposed, postponed, or weakened? In May 2023, Altman warned that OpenAI could leave Europe if it could not comply with the AI Act. Two days later, he said it had no plans to leave. Calls for regulation are easy. Accepting rules written by society is harder. Jessica Morley, Claudio Novelli and I checked the public record. We took the predictions seriously. Our research “Not Even Wrong 2: An Audit of Public AGI Prediction, 1950–2026” audits 24 widely circulated predictions about AGI and related capabilities. Open access: https://lnkd.in/gn3yGrVM Spoiler: • 19 are not even wrong: they lack adequate definitions, measurement or independent adjudication. • 2 were testable, and their deadlines have passed. Both were wrong. • 3 are testable and remain open. Predictions from commercial labs and company leaders score worst on average in our sample. And yes, Amodei is among those we audit. The audit is not about AI risks. This is the topic of another study. It's about predictions. And it proves that a claim that survives any outcome is not evidence; however, many mastheads lend it credibility. Presented as foreknowledge, this is theatre, distracting from problems already with us: • oligopoly and concentrated control over compute; • gaps and delays in binding regulation; • energy and water costs, incompletely disclosed; • unequal gains and underpaid annotators; • risks to cultural diversity and independent learning; • bias amplified in hiring, credit and justice; • surveillance and coercive state power; • AI used in military targeting in Ukraine and Gaza. These problems require timely decisions about systems and institutions already shaping our lives. Why, then, so much emphasis on AI-pocalypse instead? P.S. We are already doing an excellent job of endangering our own survival. No superintelligence required.
-
Natasha Lomas shared thisThe mere existence of watermarking for AI-generated text is proving to be a very *psychologically interesting* development as certain tech bloggers have a fit of the vapours over detectability or 'le mot juste artificiel'. More from Maximilian Henning: https://lnkd.in/eJssVJ4kEU’s push for watermarking won’t worsen AI writing, experts say | EuractivEU’s push for watermarking won’t worsen AI writing, experts say | Euractiv
-
Natasha Lomas shared thisHappy Fr(A)iday!Natasha Lomas shared thisThis is the only type of AI hot take I'm interested in on LinkedIn, frankly. Credit to creator @husk.irl for his content. If "AI is inevitable" we're kind of cooked.
-
Natasha Lomas shared thisPreachNatasha Lomas shared this“You either work with us or you don’t”. What sounds like a line from a bad gangster movie from the 1980s, is the US ambassador “warning” the EU against introducing “protectionist” rules ahead of incoming cloud and semiconductor laws – suggesting a planned tech sovereignty packagecould threaten the EU-US trade agreement as Euractive’s Théophane Hartmann reported early on. And guess what: On Tuesday, the tech sovreignty package was postponed for the third time. It was due March 25. Then April 15. Now May 27 and it’s been delayed again. A set of landmark proposals including the Cloud and AI Development Act, Chips Act 2, and a strategy for open-source software. Together, they would define what makes a cloud “sovereign”, reduce Europe’s dependency on non-EU tech infrastructure, and lay the groundwork for homegrown Europe’s digital agenda is being used as a bargaining chip. There’s one thing everyone dealing with blackmailers knows: Never give in. It will never stop. Stand up for yourselves. Free yourselves from the dependency while you still have a powerful internal market that US companies are very interested in. When will the European Commission finally stand up for Europe? You don't build sovereignty by kissing the ring!
-
Natasha Lomas reposted thisWhat Maximilian Henning writes on yesterday’s trilogue is extremely serious. With: 🔴 the Commission abandoning its honest broker role so blatantly and from up top, 🔴 one country jeopardizing the whole Council mandate from within, just to please a couple of companies, 🔴 the EPP again leveraging the far right support ➡️ the risk is torpedoing the AI Omnibus and, not less important, undermining the standardization ecosystem, which should be a strategic asset for Europe in the global context. I cannot see anything positive coming from all this, not only for AI governance, but for European democracy itself. There is still a small window to fix this. Let’s all wake up before it’s too late. Alberto Alemanno Marietje Schaake Adam Leon Smith DEng FBCS Sebastian Hallensleben 🇪🇺Paul NemitzNatasha Lomas reposted thisSocialist and Green lawmakers are taking square aim at the EPP over tonight’s failed AI talks – not only within Parliament, but also in Berlin and in the Berlaymont. The Greens are even speaking of a “German EPP coup”. Read my article at Euractiv: https://lnkd.in/eqj5ZXdQLeft-wing lawmakers blame Commission for stalled AI reform talks | EuractivLeft-wing lawmakers blame Commission for stalled AI reform talks | Euractiv
-
Natasha Lomas shared thisMy brilliant friend and ex-TC colleague showing that smart (human) precision is the best riposte to AI slurryNatasha Lomas shared thisI launched my tech newsletter last week. After more than a decade covering startups and tech in Europe, I wanted to build something more focused: a smart briefing for people who want to understand what’s really happening in tech, and see how it’s going to change the (business) world around us. So if you're a founder, an investor or an operator and want to go beyond the news headlines, subscribe to get my analysis and opinion on tech in Europe. 👉 Hypertext.fyi
-
Natasha Lomas reposted thisGoogle is the tech company the most hostile to democracy. No one else has done more to defund the media. They tend to still get a vague pass in technical circles because most computer geeks have a poor understanding of the web's economics — where the search/browsers deal defunds the entire ecosystem — and an even poorer understanding of politics and how information works in society. But they will treat any independent source of information as a bug that needs eliminating. Organising the world's information was always an authoritarian project and it's really coming into its own. Most of the news industry is terrified and traumatised. We need more people to speak up like Mike below, the industry needs to be a lot less submissive and a lot more strategic, and policymakers need to see through the bullshit and act accordingly.Natasha Lomas reposted thisThe photo below was taken at Google HQ. I was told my 24 years of experience mattered. I was told Google wanted to 'fix' the search ecosystem. 17 MONTHS later, they’ve ghosted us and our traffic is being cannibalized by their AI. The 'conversation' was a lie. Google invited me to HQ to help "fix" search. It was a pacification tactic. I’ve been publishing Mountain Weekly News for 24 years. I’ve spent two decades technically vetting gear in the trenches of the Tetons while Google was still in its infancy. I walk the walk, I test the gear, and I’ve survived every algorithm shift—until now. October 2024, Google invited me and 19 other independent creators (including Rutledge Daugette of TechRaptor) to their HQ for a "Web Creator Conversation." They told us they wanted to hear from the experts. They told us search was changing and they needed our help to get it right. It was a masterclass in smoke and mirrors. Danny Sullivan (who won't allow me to tag him...) and the search team asked us to stop "pining" them publicly on X and LinkedIn. Instead, they asked us to compile "internal documents" with our findings, data, and technical concerns. They promised these would be the "workbooks" for the engineering team. Rutledge sent 26 pages of exhaustive data. I sent an 8-page deep dive on how Google was rewarding corporate conglomerates over actual experts. The result? Silence. Ghosting. Total lack of accountability. Did anyone on your team see this Cathy Edwards? We weren't invited there to provide solutions; we were invited there to be silenced. It was a pacification tactic to shut us up while they finished building a parasitic search engine that values Reddit threads and AI-generated "slop" over two decades of primary-source expertise. The "Don't Be Evil" era is dead. Why did that get removed Jennifer Kutz Google has a responsibility to the ecosystem that feeds it. Instead, they’ve chosen a path that rewards media conglomerates that pay writers pennies to "spin" content, while independent publishers who actually invest in original photography and technical vetting are forced into layoffs and closure. If Google continues to prioritize "Zero-Click" results and corporate authority over human expertise, there will be no professional independent media left to crawl in two years. Any thoughts Pandu Nayak? Should I close a 24 year old business? Rutledge and I aren't giving up. We’ve found business models that don't rely on a broken search engine. But the open web is on life support, and the "physician" is the one pulling the plug. Check the data for yourself: any thoughts? Cyrus S. Barry Schwartz Lily Ray Glenn Gabe My report: https://lnkd.in/g4xCgg8Q 26 Page Report: https://lnkd.in/gbwPXVk4 #GoogleSearch #SEO #HCU #DigitalMarketing #IndependentPublishing #TechRaptor #MountainWeeklyNews #SearchEngineLand #GoogleHQ #AIO #ZeroClick
-
Natasha Lomas reposted thisNatasha Lomas reposted thisWith more than 100,000 attendees from over 200 countries at MWC Barcelona 2026, the energy around AI, 5G Advanced and autonomous networks was unmistakable. Our industry is clearly moving beyond connectivity into a new era where intelligence is embedded into everything — from AI‑native infrastructure and programmable networks to robotics, immersive XR and next‑generation digital platforms. Singtel was proud to contribute to this momentum — showcasing our solutions across partner booths and signing new agreements with global leaders to accelerate industry transformation. Some highlights from the week: 🔹 Robotics for real‑world deployment with AGIBOT We signed an MoU with AGIBOT, a leader in humanoid robotics. Beyond the impressive demos, the real opportunity lies in enterprise use cases across retail, logistics, manufacturing and services, where robotics, AI and connectivity converge to drive productivity and resilience. 🔹 Agentic AI with Sierra We announced a strategic partnership with Sierra to deploy agentic AI across chat and voice channels. Our pilot in Singtel Singapore’s customer care ecosystem, with AI assistant Shirley, handled over 70,000 customer cases in six weeks — resolving 73% of mobile and home troubleshooting incidents and completing 76% of roaming sign‑ups independently. And yes, she understands Singlish — because local context matters. In a fireside chat with Sierra CEO and co‑founder Bret Taylor, moderated by Natasha Lomas, we discussed why successful AI adoption isn’t just about technology — it’s about rethinking processes, empowering people and reshaping organisations. 🔹 Advancing 5G into an AI‑powered platform with Ericsson Together with Ericsson, we are accelerating 5G Advanced into a programmable, AI‑driven digital platform — strengthening Singapore’s digital economy and enabling next‑generation industry use cases. 🔹 Expanded collaboration with China Unicom Advancing cross‑border connectivity across travel eSIM, global IoT and 5G SA roaming. MWC 2026 reaffirmed that innovation thrives through collaboration — and I’m excited about what lies ahead as we build a more intelligent, resilient and sustainable digital future together. #Singtel #MWC26 #MobileWorldCongress2025 #AI #5G #5G+ #AutonomousNetworks #IoT #DigitalTransformation #DigitalTrust #EmpowerEveryGeneration
Natasha Lomas replied to a comment
6mo
B Cavello Well, yes, but they wrote: "Policymakers and advocates who are serious about children’s safety and privacy online should coalesce around a ban on behavioral advertising and on profiling young people. "
-
Natasha Lomas reacted on thisNatasha Lomas reacted on thisAnother day when I turn one of the major news programmes in the UK to find at the top of the news an earnest "expert" explaining that AI just goes wrong, there's no intention so there's no liability. We can only place our hope in the self regulatory restraint of the godlike tech companies. At the same time, every bit of language they use has been constructed to sound like these agents are perhaps small dangerous animals, completely not under the instructions and designed from scratch by the companies that plan to profit them. The presenter, usually a young woman with nice hair, looks worried. The public listen and viewing figs go up. Everyone is happy. Like everyone I know, I am intensely frustrated that the voices of sense which absolutely do exist & are occasionally even given a gasp of air time ( hello Wendy Hall ) simply cannot seem to cut through to the general public through the miasma of manipulated hype about existential AI risks and the semi-mystical loss of control. We could start with a statement that self regulation is not and has not been the status quo for several years, as both the media and the US government seem determined to tell us, and that plenty of regulation already exists in bespoke legislation both in EU , China and states like California. But even if you deny these can be enforced ( and if so we are in a de facto cold war and that should be on the table) , what has happened to basic principles of law? If we replaced AI with a malfunctioning car or dishwasher or even dog, would we be so inert? Why are basic laws regarding hacking and cybersecurity, tortious liability for failure in duty of care, misuse of data, not being explored? Yes there are doctrinal problems depending on jurisdiction but if we managed to construct a theory of corporate manslaughter in the courts I'm pretty sure we can work with this - if the will is there Meanwhile it does feel like everyone I know on this forum is a victim of a giant communal gaslighting.
-
Natasha Lomas reacted on thisNatasha Lomas reacted on thisIt's a big day for everyone here at Computer Weekly. Sixty years ago today, on 22 September 1966, IT managers across the UK received what would soon become a familiar thump landing on their desks every week – the launch issue of Computer Weekly, the world’s first weekly technology newspaper. Our 60th anniversary special issue was emailed out as a PDF – an acronym that would have flummoxed those 1966 readers. Six decades has produced a lot more acronyms along the way – a lexical journey charting some of the most incredible innovations the world has ever known. We’re proud to have been there every step of the way, the only business technology publication to have followed and championed the progress of the UK IT community for that whole time. Below is a short video looking back at that first issue - how much has changed since then, and yet, how familiar some of the stories still seem today. And here is our collection of 60th anniversary essays - 60 of them, (well, 62, but 60 feels more appropriate!) - contributed by our friends from across the industry: tech leaders, suppliers execs, politicians, analysts, inventors, innovators, experts and journalists. A unique set of personal memories recounting the human impact of technology during Computer Weekly's lifetime: https://lnkd.in/eZywbS9v Happy birthday to us and a massive thank you to everyone who has been a part of the Computer Weekly story for these last 60 years.
-
Natasha Lomas reacted on thisNatasha Lomas reacted on thisWhat's crazy is Anthropic themselves documented terrorist groups using their AI to build ballistic missiles, yet their CEO is still fixated on imaginary cyberattacks that are literally impossible. One single missile could kill more people than every cyberattack in the history of the internet. But, of course, the real risk is botnets. What if someone made a really big botnet? It's not something that's happened like 500,000 times already. It would be super serious. Just think about it. A botnet. A really big botnet. Think of all the bots and the netting it could do. Would surely bring about the end of times. It's a well know fact that if you get hacked too many times, you die. I am actually gonna lose my whole damn mind if this bubble doesn't burst soon and I have to keep listening to the most clueless people on earth opine about cybersecurity.
-
Natasha Lomas liked thisNatasha Lomas liked thisThis has been a week of exceptional BS in AI-land from "superintelligence" to "dangerous AI" (dangerous disembodied tech mind you! not dangerous companies with deeper pockets and more market power and entitlement than anyone's ever seen!!) to yet another round of wind-bloated predictions about AI's economic impact... In the middle of all of this, we thankfully do have some terrific foundational scholarship to draw on. For example, "AI auditing: The Broken Bus on the Road to AI Accountability" by Abeba Birhane, Ryan Steed, Victor Ojewale, Briana Vecchione, PhD, and Deborah Raji is excellent context to better understand what is and is *not* happening when OpenAI or Anthropic asks an external organization to review their security failures. Open-access version available here: https://lnkd.in/ggbzADFQAI auditing: The Broken Bus on the Road to AI AccountabilityAI auditing: The Broken Bus on the Road to AI Accountability
-
Natasha Lomas liked thisNatasha Lomas liked thisOh my goodness, this is a fabulous paper. It argues that measuring AI through automation capability and adoption misses what actually happens to work, and that we need to move beyond atomised tasks and look at whole workflows. Task allocation. Decision rights. Human review. Intervention. Exception handling. Recovery. And it gets better. Genuine augmentation requires what they call durable net value. If apparent productivity gains are swallowed by verification, exception handling, rework, recovery or cognitive burden, you may have just moved the work elsewhere. Remember the bottleneck always moves. They also make a point I really like: human oversight only works if the human retains the competence, time, information and authority to actually challenge the system. And the way we redesign the work can itself erode that competence over time. Amen. Finally, a paper that just comes out and says it. Been waiting ages.
-
Natasha Lomas liked thisNatasha Lomas liked thisThe Light Bulbs You Need to Get Rid of Immediately (New Research). The light bulb above your head right now may be quietly STARVING every cell in your body. Scientists just gave it a name: 21st century SCURVY. In less than one generation we deleted 95% of the natural light spectrum from human life, and almost nobody noticed. Press your finger over your phone's flashlight and you'll see a deep red glow. That's near-infrared light passing straight through your hand - the exact wavelength band that LED bulbs and energy-efficient windows have almost completely erased from the places where we spend 90% of our lives. In this video I break down one of the most unsettling new ideas: light isn't just something you see, it's something your mitochondria consume like a NUTRIENT. Remove it and blood sugar rises, energy drops, and the aging brain loses protection it evolved over billions of years to depend on. If this helped you see your living room differently, please share it with a caregiver, a parent, or anyone managing diabetes or cognitive decline.
-
Natasha Lomas reacted on thisNatasha Lomas reacted on thisWho says you can't be creative using AI generation... if you use your own creativity, of course. This is the sort of creativity using AI that hits it right on the nose. Credit: Ari Kuschnir (@arikuschnir) https://lnkd.in/e2KeRmeG
Experience & Education
Languages
-
Spanish
Professional working proficiency
-
English
Native or bilingual proficiency
Recommendations received
1 person has recommended Natasha
Join now to viewOther similar profiles
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content