The US has required Anthropic to restrict access to its most advanced models Fable 5 and Mythos 5 to foreign nationals. Anthropic disabled these models for all non-US users, setting a precedent for sovereign control of advanced AI systems.
On June 12, 2026, a U.S. government directive forces Anthropic to suspend Fable 5 and Mythos 5 for all users, citing jailbreak risk. Anthropic complies but contests, arguing a potential narrow workaround does not justify recalling a widely deployed model.
Prometheus, physical AI startup co-founded by Jeff Bezos and Vik Bajaj in late 2025, raises $12B Series B at $41B valuation. JPMorgan and Goldman Sachs participate in the round.
OVH Groupe is in exclusive negotiations to acquire Gladia, a French startup specializing in speech recognition and AI transcription. The acquisition aims to strengthen OVH's sovereign cloud offering by adding native audio processing capabilities.
Apple integrates Google's Gemini into Siri at WWDC on June 8, but not as the primary engine. The article challenges the dominant interpretation of this partnership and explores Gemini's actual role in Siri AI's architecture.
A randomized controlled trial (arXiv, April) measures the impact of learning with GPT-5 on skill retention after assistant removal. Results quantify the cognitive cost of AI dependency.
OpenAI filed a confidential S-1 with the SEC on June 9, eight days after Anthropic. Both AI companies are preparing IPOs, disclosing financial data previously hidden in private valuations.
Lille hosts the « L'IA avec nous » summit on June 12 at EuraTechnologies with 1,000+ participants and 50 speakers. The event tests positioning of a European applied AI valley, bringing together French and international stakeholders around concrete use cases.
Anthropic launched Claude Fable 5 and Claude Mythos 5 on June 9, 2026—two products built on the same underlying model but differentiated by distinct safety guardrails. This strategy reveals a segmentation approach based on security controls rather than architecture.
Gladia positions Solaria-3 as leader in production audio transcription (noisy meetings, accents, telephony). The transcription API market has shifted toward these complex use cases since 2024-2025.
Google Research presents at AISTATS 2026 a statistical test to audit machine unlearning (targeted data erasure). The method is validated on classical models but not yet on LLMs, raising questions about applicability to large language models.
Uber and Wayve pursue a fleet/software separation model for robotaxis in London (June 2026), while Waymo favors vertical integration. The two strategies diverge on value chain control.
LARA, an audit tool, evaluates RGPD and AI Act compliance of 12 major LLMs. None meet the compliance threshold. Regulatory risk shifts from integrators to data protection officers.
Nvidia and Hyundai Motor Group announce on June 8, 2026 a new cooperation phase. Nvidia offers a hybrid model: open software (Isaac) and proprietary silicon, aiming to establish a robotics platform in South Korea.
Canada lacks a binding federal regulatory framework for AI, unlike other G7 countries. The CCPA criticizes the voluntary approach and absence of enforceable compliance presumption, deemed insufficient to ensure corporate accountability.
UK's CMA imposes technical opt-out for Google's AI Overviews without establishing publisher compensation rights. This framework differs from contractual licensing based on neighboring rights.
Uber caps monthly spending at $1,500 per employee for agentic coding tools (Claude Code, Cursor) after exhausting its AI budget in four months. The measure aims to control expenses from code agents.
Helsing, a German startup, unveils the RX-1 on June 1, 2026. The European military robot relies on American inference silicon, exposing Europe's technological dependence in defense robotics despite sovereignty ambitions.
Since June 2026, European digital authorities using Qwen or DeepSeek must comply with the AI Act requiring disclosure of training data. Beijing refuses to share them, creating a major regulatory conflict between the EU and Chinese providers.
The Linux Foundation unveils DNS-AID (Agent IA Discovery), an open-source project using DNS as a discovery layer for AI agents. This decentralized approach challenges proprietary registries from hyperscalers (OpenAI, Google, Anthropic).
Anthropic grants the ENISA (European cybersecurity agency) early access to Claude Mythos, two months before the European Commission has legal leverage to require it. First European institution to receive this access.
STMicroelectronics raises its data center revenue guidance to $1 billion for 2026, despite an 89% drop in net income. The Franco-Italian chipmaker adjusts forecasts upward for the data center segment.
Starbucks withdraws NomadGo, its computer vision tool deployed across 11,000 stores after nine months. The system, validated at 99% in controlled environment, failed at production scale, highlighting the gap between sandbox testing and real-world deployment.
CNIL fines IQVIA Operations France €5M (decision SAN-2026-008, May 26, 2026) for GDPR breach. Data pseudonymization does not eliminate legal obligations if the company retains decryption keys.
Choose France 2026 displays €93 billion in announced investments, of which €45 billion are firm contractual commitments. SoftBank accounts for half the volume.
ExpGraph, an arXiv preprint from May 29, 2026, proposes a self-evolving graph memory architecture for LLM agents. The system accumulates experience through dynamic nodes and edges, enabling agents to improve performance on complex tasks without retraining.
Nvidia announces Alpamayo 2 Super 32B, positioned as the foundation of its robotaxi stack. Embedded performance metrics remain unpublished, raising operational questions for automotive R&D directors.
Four competency frameworks for LLM agents published in May 2026 without unified protocol. CTOs and DSIs must choose between incompatible skill library architectures for their multi-agent systems.
French State invested €404 million in March to acquire Bull from Atos, asserting sovereignty in high-performance computing. Foxconn establishes a factory in Angers, raising concerns about French technological independence amid global supply chains.
SoftBank commits €45 billion in France for AI data center development in an initial phase, with €75 billion potential subject to expansion. The project targets up to 5 GW capacity.
Anthropic raises $65 billion in Series H funding, reaching a $965 billion valuation. No European public funds participated in the funding round.
Nvidia invests $300M in Decart, a startup focused on world models and software optimization. Nvidia's participation aims to control an optimization layer capable of running on its chips and those of competitors.
Cigref estimates €140 billion in annual cloud and software overruns across European organizations. AI bundled into solutions ranks as the second identified cause. Half of CIOs cannot measure ROI from these integrated AI offerings.
The EU postpones to December 2027 the enforcement of obligations for high-risk AI systems in HR tools. A provisional political agreement on May 7, 2026 regarding the Digital Omnibus AI amends the timeline of regulation 2024/1689.
The fourth edition of PCAIDE (Paris Conference on AI & Digital Ethics) will take place on June 11-12, 2026 at Mines Paris. The conference returns following the 2025 edition.
Mistral AI showcases its industrial pivot at AI Now Summit (May 28, 2026) with announced partnerships with EDF, BMW, and Airbus. However, specific contract values remain undisclosed.
ByteDance is developing custom Arm and RISC-V processors to reduce inference costs for its models. The group processes 120 trillion tokens/day with Doubao and aims to reduce Nvidia GPU dependency by optimizing server infrastructure.
Anthropic granted operational access to Claude via Project Glasswing to the US Federal Reserve and Bank of England, but no EU institution has such access based on available information.
Anthropic releases Claude Opus 4.8 on May 28, 2026. The model is reportedly four times less prone to errors, with emphasis on honesty regarding its own failures.
The CIANum calls for breaking down public-private silos to build strategic commons against French and European dependence on foreign tech platforms. Goal: strengthen European digital sovereignty through structured collaboration.
GPT-4o, ChatGPT, and GPT-o3 display confidence exceeding their actual accuracy, with the gap widening on difficult tasks where they make the most mistakes. A USC/Berkeley preprint reveals growing divergence between stated confidence and real performance.
Benchmark study across 23 models showing that context compaction, the standard technique for long agent sessions, does not fix persona drift. ContextEcho evaluates this critical limitation in current systems.
Huawei unveils LogicFolding, a 3D density technology targeting 1.4 nm by 2031 without EUV equipment dependency. He Tingbo, head of Huawei's semiconductor division, announced it on May 25, 2026 at IEEE ISCAS conference in Shanghai.
The ECB convened supervised eurozone banks in late May 2026 to address cybersecurity risks related to Mythos. DORA regulation does not guarantee sovereign access to this tool for financial institutions.
EQT appointed manager of the Scaleup Europe Fund, a €5 billion fund dedicated to European technological sovereignty. No French institutional investor is a founding LP.
France ranks 2nd in declared AI maturity in Europe, yet 49% of projects remain in pilot stage and 80% have no measured ROI. Gap reveals disconnect between stated adoption and actual impact.
Anthropic leases Colossus 1, xAI's supercomputer, for $1.25B/month through May 2029 ($40B+ total). The contract caps Anthropic's access at 11% of cluster capacity, restricting the company to a fraction of available resources.
The Darcos bill on copyright and AI was removed from the French National Assembly's agenda on May 12, 2026. Without article L. 331-4-1 of the French Intellectual Property Code, French rights holders lack protection against the obligations of article 53 of the European AI Act.
Celonis acquires Ikigai Labs and launches the Celonis Context Model (CCM). MIT joins as investor in exchange for a patent license. Announced May 12, 2026 from Munich.
Île-de-France Region launches a call for proposals to design, build and operate an AI-dedicated Third Place in Paris. Public-Private Partnership (PPP) scheme aimed at creating an innovation and collaboration space.