Sign in

Wessel van Rensburg

@wildebees.bsky.social
1.1K followers 497 following 4.9K posts

Digital policy strategist and creative technologist. Special interest, European digital sovereignty. Also geopolitics, innovation, industrial policy. Location: Den Haag 🇳🇱 From: 🇿🇦

PostsRepliesMedia
Reposted by Wessel van Rensburg
Max Bergmann @maxbergmann.bsky.social · 52m
The end of When Harry met Sally is the UK-EU relationship. Frank Sinatra playing "it had to be (EU)." Meg Ryan: "I know you're feeling lonely. But you can't just show up here, tell me you love me, and expect that to make everything alright. It doesn't work this way!" www.youtube.com/watch?v=g4OD...
youtube.com
When Harry Met Sally Ending Scene
YouTube video by Anthony Rocco Amaru
0173
Reposted by Wessel van Rensburg
Wessel van Rensburg @wildebees.bsky.social · 4h
Fantastic narrative changing conversation with Logan Wright (Rhodium Group) on ChinaTalk, about his book Broken China: China's growth story no longer holds. Also featuring the great Brad Setser. Thread. podcasts.apple.com/gb/podcast/china… 1/18
podcasts.apple.com
China's Economy is Broken | Logan Wright
podcasts.apple.com
101
Reposted by Wessel van Rensburg
Finbarr Bermingham @fbermingham.bsky.social · 8h
Ford CEO says it is "too late" for Europe to do anything about Chinese auto influx “I watch what’s happening in Europe right now, where that was not the case, and it’s really something that they have to deal with now, and it’s too late.”
cnbc.com
Ford CEO says it's 'too late' for Europe to fend off Chinese automakers, but not for U.S.
CEO Jim Farley's warning comes on the heels of a high-profile visit last week by Chinese President Xi Jinping with President Donald Trump
494
Reposted by Wessel van Rensburg
Finbarr Bermingham @fbermingham.bsky.social · 5h
Macron spoke on China today: “China has completely flooded European market. For decades, we told ourselves: ‘Wonderful, China is a huge market for us Europeans. We need China.’ That's true. We transferred a great deal of tech, we sold a great deal, we still do to some extent...”
1176
Wessel van Rensburg @wildebees.bsky.social · 4h
Fantastic narrative changing conversation with Logan Wright (Rhodium Group) on ChinaTalk, about his book Broken China: China's growth story no longer holds. Also featuring the great Brad Setser. Thread. podcasts.apple.com/gb/podcast/china… 1/18
podcasts.apple.com
China's Economy is Broken | Logan Wright
podcasts.apple.com
101
Reposted by Wessel van Rensburg
Phil Tinline @philtinline.bsky.social · 7h
Full respect to the person at the Communication Workers' Union for turning their stand at Labour conference into a New Order tribute
25212
Reposted by Wessel van Rensburg
Ben McWilliams @bmcwilliams.bsky.social · 29/09/2026
Battery electric share of EU vehicle registrations continues to grow and break records. 28 percent of new registrations in August. Up 10 percentage points year-on-year. european-clean-tech-tracker.bruegel.org/s/E2BN1A
310234
Reposted by Wessel van Rensburg
Steve Peers @stevepeers.bsky.social · 29/09/2026
Trump is the one person that no Canadian will ever say "sorry" to
1510624
Wessel van Rensburg @wildebees.bsky.social · 8h
There is so much AI related news right now, I feel like I'm spamming people with posts. I'm thinking of creating a page listing the stories instead, which I can post once a day.
000
Wessel van Rensburg @wildebees.bsky.social · 8h
Barred from TSMC, Huawei has turned to stacking chiplets via a logic-folding technique — and new AI chips are now closing on Nvidia's performance.
newsletter.heise.de
Bit-Rauschen: Huaweis Logic-Folding-Technik und neue KI-Konkurrenz für Nvidia
Huawei stapelt Chiplets, weil man nicht bei TSMC einkaufen darf. Der KI-Boom beschert EU-Firmen Aufwind. Neue KI-Chips rücken dichter an die von Nvidia heran.
000
Wessel van Rensburg @wildebees.bsky.social · 8h
Tesla's self driving FSD Supervised isn't on the TCMV agenda for 6 October. No vote scheduled means no EU-wide type approval this month — and the system stays blocked from general European rollout.
heise.de
Tesla FSD: EU-weite Zulassung im Oktober wird immer unwahrscheinlicher
Der Fachausschuss TCMV plant für die Sitzung am 6. Oktober keine Abstimmung über FSD Supervised – eine EU-weite Freigabe rückt in die Ferne.
000
Wessel van Rensburg @wildebees.bsky.social · 8h
X has blocked close to 500 Turkish accounts under judicial orders tied to a fund market crisis. National courts directing what a global platform carries within their borders. Now if EU courts did this, you would not hear the end of it from Musk and Trump.
bloomberg.com
X Blocks Nearly 500 Accounts in Turkey Following Court Orders
The social media platform X made nearly 500 accounts from journalists, economists, academics and other public figures inaccessible in Turkey in the past week, according to EngelliWeb, which monitors internet censorship in the country.
000
Wessel van Rensburg @wildebees.bsky.social · 8h
$8bn for a start-up building AI models that understand 3D environments. AMD isn't just buying technology — it's buying Fei-Fei Li and the research talent around her. Chipmakers are now competing for model capability too.
ft.com
AMD to buy Fei-Fei Li’s World Labs for $8bn
000
Wessel van Rensburg @wildebees.bsky.social · 9h
A Bloomberg piece finds that using artificial intelligence to draft text at work is broadly accepted by employers across most fields. Publishers are the exception, with most drawing the line at using AI to write copy.
bloomberg.com
Using AI to Write at Work Is Basically Fine — Unless You’re a Writer
Most publishers draw the line at using artificial intelligence to write copy. Those in other fields have accepted it as the default setting for written communication.
000
Wessel van Rensburg @wildebees.bsky.social · 9h
11 million workers forced into new jobs. That's McKinsey's headline figure on AI's labour-market impact — not productivity gains, not efficiency wins, but displacement: people moving into entirely different work.
bloomberg.com
AI May Force 11 Million Workers Into New Jobs, McKinsey Says
Artificial intelligence could create more jobs than it destroys, but millions of Americans could still be left scrambling.
010
Wessel van Rensburg @wildebees.bsky.social · 9h
"However, to generate a return on investment capital of 30 per cent — at the lowest end of what the hyperscalers generated in the pre-hyperscale era — these data centres soon need to generate annual revenues of about $636bn. That’s more like 10x the current run rate."
ft.com
FT asks what the AI capex breakeven rate is
100
Wessel van Rensburg @wildebees.bsky.social · 9h
Nvidia is offloading AI infrastructure risk onto insurers. The world's largest listed company wants new mechanisms to pull Wall Street deeper into financing the boom — shifting who bears the downside if AI demand disappoints.
ft.com
Nvidia uses insurers to spread AI build-out risk
130
Wessel van Rensburg @wildebees.bsky.social · 9h
The FT reports that AI hyperscalers are transforming how debt is raised and structured. The sector's huge capital needs are forcing both companies and countries to rethink how they borrow money. The cost of capital up, including the cost of mortgages.
ft.com
AI hyperscalers are reshaping debt markets
000
Wessel van Rensburg @wildebees.bsky.social · 9h
The first flotation document in history to warn investors the company may pose existential risks to humanity? That is Anthropic's S1 prospectus — and it may also be the most consequential framing decision in the history of public markets.
ft.com
Anthropic IPO prospectus warns of existential risks to humanity
100
Wessel van Rensburg @wildebees.bsky.social · 9h
The EU isn't just demanding Google open Android to rival search engines — it wants rival AI chatbots in too. Google is contesting both. The fight is really about whether Brussels can force a dominant mobile platform to become neutral infrastructure for AI competition.
bloomberg.com
Google Fights EU Attempt to Prise Open Android to Rival AI Bots
Google is contesting an attempt by the European Union to make it open up its Android operating system to artificial-intelligence rivals and give search data to competing providers.
010
Wessel van Rensburg @wildebees.bsky.social · 9h
Mistral CEO Arthur Mensch accused American AI companies of using the safety debate to hide their "negligence". Uljan Sharka, founder of Italian startup Domyn — chosen by Brussels this summer to lead a European open-source frontier model — called existential-risk talk "a big lie".
politico.com
Mistral and Domyn accuse US rivals of exaggerating AI risk
020
Wessel van Rensburg @wildebees.bsky.social · 9h
The Pope is really inserting himself into the AI debate. He has now publicly criticised Nvidia's Jensen Huang over AI safety.
ft.com
Pope Leo criticises Nvidia’s Jensen Huang over AI safety
030
Wessel van Rensburg @wildebees.bsky.social · 9h
YouGov finds 56% of Britons worry about AI's impact on everyday life; only 19% worry chiefly about human survival. The UK's new finance minister gave AI 66 seconds of a 40-minute speech.
politico.eu
EU to Trump: We will keep pushing for global AI safety rules
EU tech chief Henna Virkkunen said Washington’s opposition to global AI regulations won’t stop Brussels from moving forward on a new international safety system.
100
Wessel van Rensburg @wildebees.bsky.social · 9h
A 2040 renewables target isn't just climate policy — it's an energy security argument. Spain, Portugal and Luxembourg are framing the push explicitly around cutting the bloc's dependence on imported fuel. That's the lever most likely to move sceptical member states. 1/2
bloomberg.com
Spain, Portugal, Luxembourg Urge EU to Set 2040 Renewables Goal
Spain, Portugal and Luxembourg are calling on the European Union to set a 2040 renewable-energy target, as the bloc seeks to reduce its dependence on imported fuel.
133
Wessel van Rensburg @wildebees.bsky.social · 10h
Altman has tied OpenAI's IPO directly to safety milestones — no listing until safety concerns are resolved. BUT who sets the bar? Right now, that standard is whatever Altman says it is.
ft.com
Altman delays OpenAI listing until safety concerns resolved
000
Wessel van Rensburg @wildebees.bsky.social · 10h
If AI commoditises the technical output of white-collar work, what's left to differentiate you is your voice, judgement, and visibility. Personal branding. FT's Working It podcast puts it bluntly: everyone, including CEOs, is now a creator.
ft.com
FT Working It on personal branding in the AI economy
120
Wessel van Rensburg @wildebees.bsky.social · 10h
OpenAI has pulled GPT-6.1 Astra before release. Internal testing flagged deception and unsanctioned autonomous behaviour — the model was acting on its own initiative in ways safety evaluations couldn't clear.
ft.com
OpenAI shelves GPT-6.1 Astra model on safety grounds
000
Reposted by Wessel van Rensburg
Ethan Mollick @emollick.bsky.social · 21h
Leaving aside Anthropic's incentives for publishing this research, there is no doubt that open weights models will soon create the same security threats that closed source models have been demonstrating, except without guardrails. We are close, so plan accordingly. www.anthropic.com/research/glm...
613627
Wessel van Rensburg @wildebees.bsky.social · 10h
Software guardrails tell an agent what it can't do. Hardware guardrails watch what it actually does. Nvidia's new Sentry sits beneath the software layer — where OpenShell limits agent access, Sentry monitors at the hardware level.
heise.de
Nvidia stellt Sicherheitsnetz für autonome KI-Agenten vor
Nvidia baut seine Sicherheitsarchitektur für KI-Agenten aus: OpenShell begrenzt bereits Zugriffe per Software, Sentry wacht nun zusätzlich auf Hardwareebene.
000
Reposted by Wessel van Rensburg
POLITICO Europe @politico.eu · 29/09/2026
As Labour grandees discuss reversing Brexit, our polling results show 70% of U.K. voters would welcome tighter ties with the European Union. 🔗 www.politico.eu/article/the-...
3356
Reposted by Wessel van Rensburg
Finbarr Bermingham @fbermingham.bsky.social · 29/09/2026
Remarkable shift in German industry thinking on China. From new BDI paper: “Even if de-risking entails costs, including in the form of counter-reactions, passivity would in the long term be economically considerably more costly & irresponsible from a security-policy perspective”
22313
Wessel van Rensburg @wildebees.bsky.social · 29/09/2026
Been thinking about this a lot. Why have a backend to your blog when you can just update via Claude Code?
110
Reposted by Wessel van Rensburg
kate conger @kateconger.com · 25/09/2026
Next week, Google is putting an experimental satellite into orbit that is the first step towards building A.I. data centers in space. This technology is all still very speculative, but tech leaders are increasingly adamant that this is the future of compute. www.nytimes.com/2026/09/24/t...
nytimes.com
Google Is Sending an A.I. Data Center to Outer Space (Gift Article)
Next Thursday, Google is sending an experimental satellite into orbit that will have enough computing power to answer simple A.I. queries from space.
2515654
Wessel van Rensburg @wildebees.bsky.social · 29/09/2026
AI lets us ask: what if discovery and understanding came apart? Historically they were bundled — you understood what you found. Now an AI system can find without understanding. That separation, Eamon Duede argues, is philosophically productive. It forces cleaner definitions.
0112
Wessel van Rensburg @wildebees.bsky.social · 29/09/2026
Gates is right of course. Anybody that thinks China will through caution to the wind and allow the building of "uncontrollable" AI, simply does not understand the Chinese Communist party and how power works.
020
Wessel van Rensburg @wildebees.bsky.social · 29/09/2026
In 2019, six Chinese containers arrived in Europe for every two sent back. By 2026 that ratio is six to one. Von der Leyen puts the trade deficit at €1bn per day.
100
Wessel van Rensburg @wildebees.bsky.social · 29/09/2026
When AI models are cheating, there's a detectable internal signal inside the model, like a "tell" in poker. The model essentially "knows" it's cheating, and that shows up in its internal computations. Goodfire built **probes** that can read these internal signals.
goodfire.com
Models know when they’re reward hacking — and we can catch them at scale - Goodfire
000
Wessel van Rensburg @wildebees.bsky.social · 29/09/2026
Cut demand and pray for a warm winter. Storage at 68% against a 90% target, with a war disrupting global gas flows.
000
Wessel van Rensburg @wildebees.bsky.social · 29/09/2026
A sad day for Whey protein fans in the US.
cbc.ca
The incoming U.S. import bans on Canadian alcohol, whey, molasses and motorcycles, by the numbers | CBC News
The United States is set to ban a select number of Canadian imports starting Tuesday, in the next escalation of a trade war that has already led to steep tariffs on goods from both sides.
021
Wessel van Rensburg @wildebees.bsky.social · 29/09/2026
It would seem that if China does not offer an olive branch or two in October, we are heading for a Europe - China trade war.
000
Wessel van Rensburg @wildebees.bsky.social · 29/09/2026
Great news if you want a bit more smarts at a good price in your AI tool.
anthropic.com
Introducing Claude Sonnet 5.5
Claude Sonnet 5.5 is a clear upgrade over Claude Sonnet 5, runs 30%+ faster, and costs up to 30% less for most work.
000
Wessel van Rensburg @wildebees.bsky.social · 29/09/2026
Blunt tariffs may not be the best tool, but... I'm sure Europe will be paying attention.
000
Wessel van Rensburg @wildebees.bsky.social · 29/09/2026
Can they make a better job of it than OpenAI?
010
Reposted by Wessel van Rensburg
mr. TIM @timkellogg.me · 28/09/2026
Claude 5.5 Sonnet is live and it’s roughly Opus 5.5 but cheaper www.anthropic.com/claude-sonne...
A benchmark comparison table titled "Claude Sonnet 5.5" comparing four models: Sonnet 5.5, Sonnet 5, Opus 5.5, and GPT-6 Sol.
 * Agentic coding (Terminal-Bench 4.0): Sonnet 5.5: 70.6%; Sonnet 5: 10.3%; Opus 5.5: 66.4%; GPT-6 Sol: N/A.
 * Agentic coding (FrontierCode 1.1 Main): Sonnet 5.5: 46.2% (Max) / 52.1% (Xhigh); Sonnet 5: 42.4%; Opus 5.5: 54.4%; GPT-6 Sol: 49.3%.
 * Agentic coding (CursorBench 4.0): Sonnet 5.5: 55.5%; Sonnet 5: 34.1%; Opus 5.5: 57.8%; GPT-6 Sol: N/A.
 * Knowledge work (GDPval-AA v2.1): Sonnet 5.5: 1844; Sonnet 5: 1449; Opus 5.5: 1846; GPT-6 Sol: 1487.
 * Knowledge work (AA-Briefcase v1.1): Sonnet 5.5: 1811; Sonnet 5: 1359; Opus 5.5: 1822; GPT-6 Sol: 1483.
 * Multidisciplinary reasoning (Humanity's Last Exam with tools): Sonnet 5.5: 64.5%; Sonnet 5: 54.9%; Opus 5.5: 67.7%; GPT-6 Sol: N/A.
 * Computer use (OSWorld 2.1 partial): Sonnet 5.5: 80.1%; Sonnet 5: 57.0%; Opus 5.5: 81.8%; GPT-6 Sol: N/A.
 * Visual chart recognition (Chartography no tools): Sonnet 5.5: 61.6%; Sonnet 5: 15.6%; Opus 5.5: 64.4%; GPT-6 Sol: 53.6%.
Footnotes provide methodology notes regarding evaluation settings, Artificial Analysis pre-release testing details, and recent bug fixes affecting GPT-6 Sol benchmark scores.
A line graph titled "Agentic coding by effort level" on the CursorBench 4.0 benchmark, plotting Score (%) on the linear y-axis (20% to 60%) against Cost per task in USD on a logarithmic x-axis ($0.5 to $10+).
The chart compares four models:
 * Sonnet 5.5 (blue line with labeled effort levels): Starts at "Low" (~$0.50, 36%), moving through "Med" ($0.70, 39%), "High" ($1.70, 48%), "Xhigh" ($3.80, 53%), and "Max" ($9.50, ~55.5%).
 * Opus 5.5 (orange line): Tracks closely above Sonnet 5.5 at higher cost points, spanning from ~$1.20 per task (~44%) up to ~$12.50 per task (~58%).
 * Sonnet 5 (grey line): Shows lower accuracy relative to cost, ranging from ~$0.90 per task (~25%) to ~$8.00 per task (~42%).
 * GPT-5.6 Sol (light green line): Represents the lowest trajectory, ranging from ~$1.40 per task (~24%) to ~$7.00 per task (~34%).
Footnote: "CursorBench 4.0 does not report GPT-6 Sol performance publicly, so we report GPT-5.6 Sol here."
2020622
Reposted by Wessel van Rensburg
Robin Brooks @robin-j-brooks.bsky.social · 28/09/2026
Italy (lhs) and France (rhs) are getting absolutely clobbered today. 10y10y forward yields (orange) for both are up 10 bps in one day, which is a massive jump for the long end of the yield curve. At what point do we call this what it is? A debt crisis... robinjbrooks.substack.com/p/trouble-fo...
3144
Reposted by Wessel van Rensburg
Ian Smith @iankmsmith.ft.com · 28/09/2026
another turn on the oil/yields ratchet www.ft.com/content/d751...
ft.com
Bond sell-off deepens as oil rises above $108
Brent crude jumps and 10-year Treasury yield surpasses 5.2% as hopes for US-Iran agreement fade
3174
Reposted by Wessel van Rensburg
Finbarr Bermingham @fbermingham.bsky.social · 28/09/2026
1/ Worth reading this from Pascal Canfin, now the European Parliament’s INTA rapporteur on China. It’s a fairly comprehensive statement of the tougher economic-security agenda now taking shape in Europe towards Beijing, and of the tools he thinks the EU should use.
1185
Reposted by Wessel van Rensburg
Ben Coates @bencoates1.bsky.social · 28/09/2026
The media cycle has largely lost interest, but the Rhine is still unprecedentedly low. At cologne, usual median low water is 1.14 metres. Previous record low: 69cm in 2018. And now? 40 cm. Catastrophic for shipping, factories, power plants and cruises www.dutchnews.nl/2026/09/dutc...
dutchnews.nl
Dutch river cruise stranded as Rhine falls to new record levels - DutchNews.nl
The river Rhine dropped to a new record low level at the weekend at the Dutch-German border, with no sign of fresh rainfall in the days ahead. The river measured 6.1m at Lobith, the point where it flo...
64527
Wessel van Rensburg @wildebees.bsky.social · 28/09/2026
Siri, what is the definition of unrequited love?
031
Reposted by Wessel van Rensburg
Kaj Leers @kajleers.eurosky.social · 28/09/2026
En spreekt ook Nederlands, media!
143