<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:wfw="http://wellformedweb.org/CommentAPI/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:sy="http://purl.org/rss/1.0/modules/syndication/" xmlns:slash="http://purl.org/rss/1.0/modules/slash/" xmlns:media="http://search.yahoo.com/mrss/" xmlns:georss="http://www.georss.org/georss" >
    <channel>
        <atom:link href="https://www.actuia.com/en/feed" rel="self" type="application/rss+xml" />
        <title><![CDATA[ActuIA]]></title>
        <link><![CDATA[https://www.actuia.com/en/feed]]></link>
        <description><![CDATA[AI news & analysis from Europe]]></description>
        <language>en-US</language>
        <pubDate>Thu, 19 Nov 2026 18:00:00 +0000</pubDate>

                    <item>
                <title><![CDATA[Microsoft launches initiative to make AI agents independently verifiable]]></title>
                <link>https://www.actuia.com/en/news/microsoft-launches-initiative-to-make-ai-agents-independently-verifiable/</link>
                <description><![CDATA[Microsoft announced at Ignite a transparency initiative bringing together more than 40 organisations to enable independent verification of AI agents. The company also put a risk score and containment controls into preview, while acknowledging that they do not reveal the behaviour of the underlying model.]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/microsoft-launches-initiative-to-make-ai-agents-independently-verifiable/</guid>
                <pubDate>Thu, 19 Nov 2026 18:00:00 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11600-gkslkdxs-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ Microsoft launches initiative to make AI agents independently verifiable ]]>
</media:description>
</media:content>
                            </item>
                    <item>
                <title><![CDATA[ANS standard makes domain names the identity for AI agents]]></title>
                <link>https://www.actuia.com/en/news/ans-standard-makes-domain-names-the-identity-for-ai-agents/</link>
                <description><![CDATA[The Agent Name Service (ANS) standard, under Linux Foundation governance since November 2026, ties an AI agent's identity to a domain name, certificates and a public log. Its seven specifications remain drafts, and the committee is dominated by GoDaddy, which holds three of the five seats.]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/ans-standard-makes-domain-names-the-identity-for-ai-agents/</guid>
                <pubDate>Thu, 19 Nov 2026 16:00:00 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11597-bav25t5k-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ ANS standard makes domain names the identity for AI agents ]]>
</media:description>
</media:content>
                            </item>
                    <item>
                <title><![CDATA[Cohere leans on HCLTech for sovereign AI in Germany]]></title>
                <link>https://www.actuia.com/en/news/cohere-leans-on-hcltech-for-sovereign-ai-in-germany/</link>
                <description><![CDATA[The Indian IT services group HCLTech and the Canadian publisher Cohere announced on 17 November 2026, in a release dated from Noida, a partnership to design and deliver "sovereign" AI solutions to ent...]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/cohere-leans-on-hcltech-for-sovereign-ai-in-germany/</guid>
                <pubDate>Thu, 19 Nov 2026 12:30:00 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11596-kjsomdlm-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ Cohere leans on HCLTech for sovereign AI in Germany ]]>
</media:description>
</media:content>
                            </item>
                    <item>
                <title><![CDATA[Europe places four supercomputers in global top 10]]></title>
                <link>https://www.actuia.com/en/news/europe-places-four-supercomputers-in-global-top-10/</link>
                <description><![CDATA[JUPITER Booster, in Jülich, is fifth in the TOP500 with 1 exaflop. It is the only European machine at this level, and HPC7, HPC6 and Alps complete the European presence in the top 10. According to ActuIA's count of the list, Europe has 161 systems for 29% of the power. The United States has 163 for 38%. The first French machine, CEA-HE, is 32nd. The French KAIROS and ROMEO-2025 lead the Green500, which measures computing per watt. Alice Recoque, Europe's second exascale machine, is due to start in the second half of 2027.]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/europe-places-four-supercomputers-in-global-top-10/</guid>
                <pubDate>Thu, 19 Nov 2026 08:00:00 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11594-stk5nnir-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ Europe places four supercomputers in global top 10 ]]>
</media:description>
</media:content>
                            </item>
                    <item>
                <title><![CDATA[Four in ten French employers use or test AI]]></title>
                <link>https://www.actuia.com/en/news/four-in-ten-french-employers-use-or-test-ai/</link>
                <description><![CDATA[France Travail surveyed over 16,000 employers on their use of AI. 38% use or test it, with a wide gap depending on establishment size. 6% think it could compete with humans in their main occupation within five years. Among employers using AI, 70% expect generative AI skills at hiring and 22% have trained their teams.]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/four-in-ten-french-employers-use-or-test-ai/</guid>
                <pubDate>Thu, 19 Nov 2026 06:30:00 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11587-ro25drjg-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ Four in ten French employers use or test AI ]]>
</media:description>
</media:content>
                            </item>
                    <item>
                <title><![CDATA[Paris Call leaves developers the power to stop their own AI]]></title>
                <link>https://www.actuia.com/en/news/paris-call-leaves-developers-the-power-to-stop-their-own-ai/</link>
                <description><![CDATA[The Paris Call for AI safety, published by the Élysée on 9 November 2026, brought together 18 states, 10 companies and 11 organisations on 18 November. Developers commit to mechanisms to stop their models, including during training and internal use, but only for models they control: copies of open-weight models are excluded, and no authority can order a stop. The text provides for no sanctions or deadlines. Google, Meta and xAI, signatories on 29 September of a White House agreement without a stop clause, are absent, as are the United States and China.]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/paris-call-leaves-developers-the-power-to-stop-their-own-ai/</guid>
                <pubDate>Wed, 18 Nov 2026 09:00:00 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11553-wsewljqx-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ Paris Call leaves developers the power to stop their own AI ]]>
</media:description>
</media:content>
                            </item>
                    <item>
                <title><![CDATA[French AI startup H Company puts agents on AWS European Sovereign Cloud]]></title>
                <link>https://www.actuia.com/en/news/french-ai-startup-h-company-puts-agents-on-aws-european-sovereign-cloud/</link>
                <description><![CDATA[H Company makes its AI agent platform available on the AWS European Sovereign Cloud. Inference for European customers who request it runs within the Union, on Nvidia-managed instances; model training remains on AWS global infrastructure. Customer content and metadata stay in the EU. An unnamed European financial services company will be the first user.]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/french-ai-startup-h-company-puts-agents-on-aws-european-sovereign-cloud/</guid>
                <pubDate>Wed, 18 Nov 2026 07:00:00 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11578-ggi77iql-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ French AI startup H Company puts agents on AWS European Sovereign Cloud ]]>
</media:description>
</media:content>
                            </item>
                    <item>
                <title><![CDATA[French startup Remanence raises $30m to train models on AI agent traces]]></title>
                <link>https://www.actuia.com/en/news/french-startup-remanence-raises-30m-to-train-models-on-ai-agent-traces/</link>
                <description><![CDATA[Remanence, a Paris startup founded by Pierre-Louis Cedoz, Hamza Benchekroun and Axel Moyal, all formerly of H Company, raises $30 million in seed funding. The round is led by EQT Ventures and FirstMark. Its platform converts AI agent execution traces into training environments to specialise open-weight models, with the customer retaining ownership. The funds will go towards research, a GPU cluster and hiring in Paris.]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/french-startup-remanence-raises-30m-to-train-models-on-ai-agent-traces/</guid>
                <pubDate>Tue, 17 Nov 2026 11:14:57 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11577-9libnyt2-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ French startup Remanence raises $30m to train models on AI agent traces ]]>
</media:description>
</media:content>
                            </item>
                    <item>
                <title><![CDATA[Sopra Steria Foundation launches €100,000 prize on AI and learning]]></title>
                <link>https://www.actuia.com/en/news/sopra-steria-foundation-launches-eur100000-prize-on-ai-and-learning/</link>
                <description><![CDATA[The Sopra Steria-Institut de France Foundation is endowing its 2027 Grand Prize for Research with €100,000, dedicated to the effects of AI on young people's learning. It rewards work already underway, conducted in a non-profit European research organisation. The call runs until 21 January 2027 on the Institut de France platform, which will announce the jury composition.]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/sopra-steria-foundation-launches-eur100000-prize-on-ai-and-learning/</guid>
                <pubDate>Tue, 17 Nov 2026 09:38:10 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11575-0gw0g42k-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ Sopra Steria Foundation launches €100,000 prize on AI and learning ]]>
</media:description>
</media:content>
                            </item>
                    <item>
                <title><![CDATA[OVHcloud unveils Olala mini, its first LLM: speeds, no scores]]></title>
                <link>https://www.actuia.com/en/news/ovhcloud-unveils-olala-mini-its-first-llm-speeds-no-scores/</link>
                <description><![CDATA[OVHcloud announced Olala mini 7A1B on 17 November 2026, an open model with 7 billion parameters, 1 billion active, due Thursday 19 November on Hugging Face under Apache 2.0. The announcement contains no evaluation results; the advertised 8 GB of memory is less than the 14 GB the weights occupy in BF16. The architecture comes from Dragon LLM, acquired in March, training relied on the equivalent of €7 million of public compute, and the only cited customer, Qwant, belongs to a company controlled by the Klaba brothers.]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/ovhcloud-unveils-olala-mini-its-first-llm-speeds-no-scores/</guid>
                <pubDate>Tue, 17 Nov 2026 07:46:15 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11569-jxy66avj-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ OVHcloud unveils Olala mini, its first LLM: speeds, no scores ]]>
</media:description>
</media:content>
                            </item>
                    <item>
                <title><![CDATA[Anthropic: $900bn for how many months' lead?]]></title>
                <link>https://www.actuia.com/en/news/anthropic-900bn-for-how-many-months-lead/</link>
                <description><![CDATA[Anthropic sets its IPO price on 18 November, with $900.9 billion in compute commitments, of which $70.7 billion due within a year. Epoch AI measures an average four-month lag for the best open models, the UK AISI four to seven months in cybersecurity; on Artificial Analysis's index, 60% based on secret tests, the gap reaches 19 points, which the body does not convert into months. METR estimates Anthropic's internal models are progressing 1.9 times faster since October 2025, with strong caveats. The prospectus cites among its risks possible pressure from open models on its prices, and no public Anthropic work on world models was found.]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/anthropic-900bn-for-how-many-months-lead/</guid>
                <pubDate>Tue, 17 Nov 2026 06:30:00 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11549-hmtyts1t-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ Anthropic: $900bn for how many months&#039; lead? ]]>
</media:description>
</media:content>
                            </item>
                    <item>
                <title><![CDATA[France's NIS 2 law: what is locked in before the 17 November vote]]></title>
                <link>https://www.actuia.com/en/news/frances-nis-2-law-what-is-locked-in-before-the-17-november-vote/</link>
                <description><![CDATA[More than 15,000 organisations will come under ANSSI supervision, up from 500 today. Registration, incident reporting within 24 hours, 72 hours and one month, the role of management bodies and fines of up to €10 million or 2% of global turnover come from the directive and will not change. Local hosting, priority to qualified providers and the ANSSI-CNIL single window, voted against the government's advice, depend on the joint committee. The text sets no compliance deadline.]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/frances-nis-2-law-what-is-locked-in-before-the-17-november-vote/</guid>
                <pubDate>Mon, 16 Nov 2026 13:30:00 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11560-o4mqmmvc-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ France&#039;s NIS 2 law: what is locked in before the 17 November vote ]]>
</media:description>
</media:content>
                            </item>
                    <item>
                <title><![CDATA[Canada gets seat on EuroHPC supercomputer board]]></title>
                <link>https://www.actuia.com/en/news/canada-gets-seat-on-eurohpc-supercomputer-board/</link>
                <description><![CDATA[Canada becomes the 39th participating state of the EuroHPC Joint Undertaking and gains a seat on the board that decides supercomputer purchases and European AI factories. Access conditions and Canada's contribution were not specified.]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/canada-gets-seat-on-eurohpc-supercomputer-board/</guid>
                <pubDate>Mon, 02 Nov 2026 09:28:00 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11542-rvk3kqkv-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ Canada gets seat on EuroHPC supercomputer board ]]>
</media:description>
</media:content>
                            </item>
                    <item>
                <title><![CDATA[ActuIA opens sector-by-sector AI news review]]></title>
                <link>https://www.actuia.com/en/news/actuia-opens-sector-by-sector-ai-news-review/</link>
                <description><![CDATA[&nbsp;As of 2 November 2026, ActuIA has published 1,342 briefs on artificial intelligence over the past thirty days, around 45 per day, drawn from 695 different sources. News that is extremely difficu...]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/actuia-opens-sector-by-sector-ai-news-review/</guid>
                <pubDate>Mon, 02 Nov 2026 06:30:00 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11539-nrjqwhd1-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ ActuIA opens sector-by-sector AI news review ]]>
</media:description>
</media:content>
                            </item>
                    <item>
                <title><![CDATA[Mistral Large 4 drops from 38 to 25 on new Artificial Analysis index]]></title>
                <link>https://www.actuia.com/en/news/mistral-large-4-drops-from-38-to-25-on-new-artificial-analysis-index/</link>
                <description><![CDATA[Version 5 of Artificial Analysis' index, published on 27 October 2026, is based 60% on private evaluations or secret answers. Twelve models with published or announced weights lose 12 to 16 points, while Claude Opus 5.5 remains at 58. The gap between the best closed model and the best published weights widens from 12 to 26 points, and the 3 to 12 point figure highlighted by Artificial Analysis includes Kimi K3.1 Pro, whose weights are not published. Mistral Large 4 falls from 38 to 25, tied with Muse Spark Open, ahead of its weight release estimated for 31 October.]]></description>
                <author><![CDATA[Stephane Nachez]]></author>
                
                <guid isPermaLink="true">https://www.actuia.com/en/news/mistral-large-4-drops-from-38-to-25-on-new-artificial-analysis-index/</guid>
                <pubDate>Wed, 28 Oct 2026 21:20:39 +0000</pubDate>
                <media:content url="https://www.actuia.com/storage/uploads/generated/11525-zf8yq9eb-1200.jpg" type="image/jpeg" medium="image">
<media:description type="plain">
<![CDATA[ Mistral Large 4 drops from 38 to 25 on new Artificial Analysis index ]]>
</media:description>
</media:content>
                            </item>
            </channel>
</rss>
