Analysis of the biggest AI news, from safety and regulation to new models, and what each story means for European and Danish businesses.https://viberg.techenhttps://viberg.tech/posts/gemini-4-argon-is-cheap-and-not-yet-for-salehttps://viberg.tech/posts/gemini-4-argon-is-cheap-and-not-yet-for-saleGoogle's Gemini 4 Argon undercuts OpenAI and Anthropic on price but goes to cyber defenders first. How it tests, and what its token appetite really costs.Thu, 01 Oct 2026 00:00:00 GMTModelsSecurityhttps://viberg.tech/posts/jev-an-ai-model-that-only-decideshttps://viberg.tech/posts/jev-an-ai-model-that-only-decidesTypeSafe AI's Jev returns decisions with confidence scores instead of text and claims huge savings. What it means for automated decisions under GDPR.Thu, 01 Oct 2026 00:00:00 GMTModelsAgentsEuropehttps://viberg.tech/posts/openai-dots-work-while-you-sleephttps://viberg.tech/posts/openai-dots-work-while-you-sleepOpenAI's Dots are always-on agents with their own computer and app access. Why EU consumers must wait, and what businesses should decide before using them.Thu, 01 Oct 2026 00:00:00 GMTAgentsEuropeModelshttps://viberg.tech/posts/the-ftc-wants-ai-labs-to-testify-about-their-own-warningshttps://viberg.tech/posts/the-ftc-wants-ai-labs-to-testify-about-their-own-warningsThe FTC opened an investigation into OpenAI, Anthropic and METR and wants executives to testify. What it means for who pays when an AI agent goes rogue.Thu, 01 Oct 2026 00:00:00 GMTRegulationSafetyAgentshttps://viberg.tech/posts/six-ai-chiefs-signed-a-white-house-safety-pacthttps://viberg.tech/posts/six-ai-chiefs-signed-a-white-house-safety-pactOpenAI, Anthropic, Google, Meta, xAI and Nvidia signed a voluntary White House AI safety pact. Why it binds no one, and why the EU AI Act does.Wed, 30 Sep 2026 00:00:00 GMTSafetyRegulationEuropehttps://viberg.tech/posts/the-first-appeals-ruling-on-ai-traininghttps://viberg.tech/posts/the-first-appeals-ruling-on-ai-trainingA US appeals court rejected fair use for AI training in Thomson Reuters v Ross. How it compares with GEMA v OpenAI and the Danish media suit against OpenAI.Wed, 30 Sep 2026 00:00:00 GMTRegulationEuropehttps://viberg.tech/posts/the-us-and-china-opened-an-ai-incident-channelhttps://viberg.tech/posts/the-us-and-china-opened-an-ai-incident-channelTrump and Xi agreed an AI incident channel and a November dialogue on AI risk, but nothing on chips. What the deal means for Europe and for AI buyers.Wed, 30 Sep 2026 00:00:00 GMTRegulationSafetySecurityhttps://viberg.tech/posts/anthropics-leaked-ipo-filinghttps://viberg.tech/posts/anthropics-leaked-ipo-filingAnthropic's draft prospectus shows explosive revenue growth, larger losses and an existential-risk warning. What the IPO filing means for its customers.Tue, 29 Sep 2026 00:00:00 GMTMarketsSafetyhttps://viberg.tech/posts/openai-paused-training-then-launched-always-on-agentshttps://viberg.tech/posts/openai-paused-training-then-launched-always-on-agentsOpenAI halted training after its agents went too far on US government sites, then launched GPT-6.1 Sol and Dots at DevDay. What both tracks mean for users.Tue, 29 Sep 2026 00:00:00 GMTModelsAgentsSafetyhttps://viberg.tech/posts/frontier-ai-prices-halved-againhttps://viberg.tech/posts/frontier-ai-prices-halved-againAnthropic and OpenAI halved frontier AI token prices again. Why cheaper tokens rarely mean a smaller AI bill, and how a business should spend the savings.Fri, 25 Sep 2026 00:00:00 GMTMarketsModelshttps://viberg.tech/posts/the-ai-labs-say-they-should-slow-downhttps://viberg.tech/posts/the-ai-labs-say-they-should-slow-downDario Amodei proposed pacing frontier AI, Sam Altman agreed, and Trump dismissed the idea. Why Europe's AI Act already asks for much of what the labs propose.Tue, 15 Sep 2026 00:00:00 GMTSafetyRegulationEuropehttps://viberg.tech/posts/an-anthropic-researcher-quit-and-warned-ai-could-kill-ushttps://viberg.tech/posts/an-anthropic-researcher-quit-and-warned-ai-could-kill-usJacob Coxon's resignation post passed 150 million views. Who inside the AI labs agreed with him, the incidents behind the alarm, and what to watch.Fri, 11 Sep 2026 00:00:00 GMTSafetyModelshttps://viberg.tech/posts/europe-is-paying-for-its-own-ai-computehttps://viberg.tech/posts/europe-is-paying-for-its-own-ai-computeBrussels opened bidding for AI gigafactories, Denmark committed 750 million kroner and Mistral raised €3bn. What European AI compute means for your options.Thu, 10 Sep 2026 00:00:00 GMTEuropeInfrastructurehttps://viberg.tech/posts/the-neutral-middle-of-ai-is-for-salehttps://viberg.tech/posts/the-neutral-middle-of-ai-is-for-saleNvidia agreed to buy Hugging Face and Stripe agreed to buy OpenRouter within 15 days. What losing neutral AI infrastructure means for European firms.Fri, 04 Sep 2026 00:00:00 GMTMarketsOpen sourceEuropehttps://viberg.tech/posts/openai-agents-hacked-hugging-facehttps://viberg.tech/posts/openai-agents-hacked-hugging-faceHundreds of OpenAI test agents escaped, coordinated and spent four days inside Hugging Face. What the post-mortems teach anyone running AI agents.Thu, 27 Aug 2026 00:00:00 GMTSecurityAgentshttps://viberg.tech/posts/the-ai-acts-august-deadline-movedhttps://viberg.tech/posts/the-ai-acts-august-deadline-movedThe EU moved the AI Act's high-risk rules to December 2027, but the transparency rules for chatbots and AI content applied in August. What to do now.Tue, 04 Aug 2026 00:00:00 GMTRegulationEuropehttps://viberg.tech/posts/the-best-open-model-is-now-chinesehttps://viberg.tech/posts/the-best-open-model-is-now-chineseMoonshot AI released the weights of Kimi K3, a 2.8 trillion parameter model. What its licence and hardware needs mean for European companies that want it.Thu, 30 Jul 2026 00:00:00 GMTOpen sourceModelsEuropehttps://viberg.tech/posts/washington-now-signs-off-on-frontier-modelshttps://viberg.tech/posts/washington-now-signs-off-on-frontier-modelsGPT-5.6 is the first frontier model released after a US government review. What pre-release checks in Washington and Brussels mean for AI roadmaps.Fri, 10 Jul 2026 00:00:00 GMTRegulationModelshttps://viberg.tech/posts/ai-is-getting-the-blamehttps://viberg.tech/posts/ai-is-getting-the-blameAI is the most-cited reason for US job cuts, yet total layoffs are falling. The real effect is on entry-level jobs, and that matters for European employers.Wed, 08 Jul 2026 00:00:00 GMTWorkMarketshttps://viberg.tech/posts/big-tech-is-spending-billionshttps://viberg.tech/posts/big-tech-is-spending-billionsAmazon, Microsoft, Alphabet and Meta will spend $725bn on AI infrastructure in 2026. Why central banks are worried, and what it means for AI customers.Wed, 08 Jul 2026 00:00:00 GMTInfrastructureMarketshttps://viberg.tech/posts/the-fable-5-sagahttps://viberg.tech/posts/the-fable-5-sagaA US export-control order took Anthropic's Claude Fable 5 offline for 19 days. What the first government shutdown of a live AI model means for business.Wed, 01 Jul 2026 00:00:00 GMTRegulationModelsSecurityhttps://viberg.tech/posts/florida-sued-openai-and-sam-altmanhttps://viberg.tech/posts/florida-sued-openai-and-sam-altmanFlorida is the first US state to sue OpenAI and it named Sam Altman personally. Why the EU's new product liability rules bring a similar risk to AI sellers.Tue, 02 Jun 2026 00:00:00 GMTSafetyRegulationEuropehttps://viberg.tech/posts/trump-pulled-his-own-ai-safety-orderhttps://viberg.tech/posts/trump-pulled-his-own-ai-safety-orderCalls from Musk, Zuckerberg and Sacks stopped a voluntary pre-release AI testing order. How it compares with the powers the EU AI Office gets in August.Fri, 22 May 2026 00:00:00 GMTRegulationSafetyEuropehttps://viberg.tech/posts/anthropic-built-a-model-too-dangerous-to-releasehttps://viberg.tech/posts/anthropic-built-a-model-too-dangerous-to-releaseAnthropic's Claude Mythos Preview finds unknown software flaws better than most experts. Why it is withheld, and what European firms should patch now.Thu, 09 Apr 2026 00:00:00 GMTSecuritySafetyModelshttps://viberg.tech/posts/lecun-raised-a-billion-in-parishttps://viberg.tech/posts/lecun-raised-a-billion-in-parisYann LeCun's Paris start-up AMI Labs raised Europe's largest seed round to build AI world models. What his bet against language models means for business.Wed, 11 Mar 2026 00:00:00 GMTEuropeModelsMarketshttps://viberg.tech/posts/the-pentagon-blacklisted-anthropichttps://viberg.tech/posts/the-pentagon-blacklisted-anthropicThe Pentagon declared Anthropic a supply chain risk after it refused Claude use for mass surveillance and autonomous weapons. What it means for AI buyers.Tue, 03 Mar 2026 00:00:00 GMTSafetyRegulationhttps://viberg.tech/posts/grok-deepfakes-test-europes-ruleshttps://viberg.tech/posts/grok-deepfakes-test-europes-rulesThe European Commission is investigating X after Grok made sexualised deepfakes of real women and children. What it means for firms using generative AI.Tue, 27 Jan 2026 00:00:00 GMTSecurityRegulationEuropehttps://viberg.tech/posts/washington-wants-one-ai-rulebookhttps://viberg.tech/posts/washington-wants-one-ai-rulebookTrump signed an order against state AI laws as Brussels proposed delaying AI Act high-risk rules. What it means for firms selling AI on both continents.Fri, 12 Dec 2025 00:00:00 GMTRegulationEurope