Revue de presse générée automatiquement avec
Profil généré automatiquement
Pulsar
Un individu à l'aise avec la technologie, fortement intéressé par l'Intelligence Artificielle et les avancées technologiques, avec un focus sur leur impact sur l'avenir du travail et la transformation numérique, à la recherche d'informations sur les tendances innovantes et les applications stratégiques.
Artificial Intelligence (30%)Technology advancements (30%)Future of work (20%)Digital transformation (20%)
Vous souhaitez recevoir chaque jour la revue de presse de ce profil ?
Embodied AI, GPT‑5.1 Leap, AI‑Driven Workflows, and Regulatory Scrutiny...
Mardi 9 décembre 2025 à 18:08
Artificial Intelligence Frontiers
Embodied AI poised to reshape manufacturing
China Daily outlines how “embodied intelligence” – robots equipped with sensors, actuators and large‑scale models – could soon outpace traditional “disembodied” generative AI, giving countries like China, Germany and Japan a decisive productivity edge. The analysis stresses that policy investment in integrated robotics will be a key lever for industrial growth.
China Daily
Skild AI eyes $14 bn valuation for robot‑ready foundation models
The Information reports that Skild AI is negotiating a $1 bn raise led by SoftBank and Nvidia, valuing the two‑year‑old startup at roughly $14 bn. Its hardware‑agnostic foundation model is designed to plug into a wide range of robots, promising faster task customization across logistics and service sectors. TechCrunch corroborates the focus on a universal model that can be fine‑tuned for diverse applications.
The Information
TechCrunch
OpenAI rolls out faster, more conversational GPT‑5.1 series
InfoQ details the three new GPT‑5.1 variants: Instant for smoother instruction following, Thinking for accelerated reasoning, and Codex‑Max for long‑running coding jobs. The upgrades cut latency by up to 30 % while expanding context windows, a move that could accelerate enterprise adoption in real‑time decision support.
InfoQ
Linux Foundation launches Agentic AI Foundation to standardise AI agents
SD Times and TechCrunch note the formation of the Agentic AI Foundation (AAIF), which now hosts Anthropic’s Model Context Protocol, Block’s Goose framework, and OpenAI’s AGENTS.md. Backed by a roster of industry giants, the foundation aims to keep agentic AI infrastructure open, interoperable and less prone to vendor lock‑in.
SD Times
TechCrunch
Claude Code brings AI‑assisted coding to Slack workflows
Computer World and The Verge describe Anthropic’s beta release of Claude Code inside Slack, where a simple “@Claude” tag lets developers surfacing bugs, retrieve repository context and even generate patches on the fly. TechCrunch highlights the potential to reshape collaborative software development by collapsing the search‑write‑review loop into a single conversational thread.
Computer World
The Verge
Technology Advancements Shaping Industry
Marvell’s $5.5 bn Celestial AI buy bolsters AI‑centric data‑center hardware
Tom’s Hardware confirms Marvell’s acquisition of Celestial AI, positioning the chipmaker to deliver next‑generation optical interconnects tailored for AI workloads, a critical upgrade as model sizes continue to explode. The deal signals intensified consolidation in the AI‑hardware supply chain.
Tom's Hardware
Global RAM shortage accelerates as AI data‑centers hoard memory
The Verge warns that memory manufacturers are prioritising AI data‑center orders, inflating DRAM prices and threatening the broader PC and smartphone markets. The article cites IDC’s Jeff Janukowicz, who predicts sustained pressure on memory supply as AI models consume ever‑larger parameter sets.
The Verge
Google prepares AI‑powered smart glasses for a 2026 launch
TechCrunch and MacRumors reveal that Google will debut two AI‑enhanced glasses in 2026: a minimalist voice‑only pair and a second version with an in‑lens display for real‑time translations and navigation. Partnering with Samsung, Warby Parker and Gentle Monster, the devices will run on Android XR and stream AI inference to a connected smartphone.
TechCrunch
MacRumors
Chrome’s Gemini agent gains a “user‑alignment critic” to block prompt‑injection attacks
The Register explains Google’s new dual‑model architecture for its Gemini‑powered browsing agent, where a secondary model reviews proposed actions before execution, mitigating the rampant prompt‑injection vulnerability that OWASP flagged in 73 % of productions. This layered defence aims to protect credentials and financial transactions in the browser.
Computer World
Future of Work Transformations
ChatGPT Enterprise saves workers up to an hour a day, study shows
ZDNet cites OpenAI’s “State of Enterprise AI” report, finding that ChatGPT users across sectors trim 40–60 minutes of routine work daily, primarily through automated drafting, data summarisation and code assistance. The productivity boost underscores AI’s shift from consumer novelty to a core corporate efficiency tool.
Zdnet
Empromptu promises no‑code AI app creation from simple prompts
TechCrunch reports that startup Empromptu raised $2 M to let users describe a desired web app in plain language and watch an AI generate the full HTML/JS stack. If the claim holds, such platforms could democratise app development and reduce reliance on traditional software engineering pipelines.
TechCrunch
Digital Transformation & Regulation
EU antitrust probe targets Google’s AI‑generated search summaries
TechCrunch details the European Commission’s investigation into whether Google’s AI‑driven search snippets unfairly leverage proprietary content without adequate compensation to publishers, a case that could reshape data‑use norms for generative services. The probe arrives amid broader scrutiny of big‑tech AI practices.
TechCrunch
European Commission examines Google’s use of web and YouTube data to train Gemini
Computer World reports that regulators have opened an antitrust case over Google’s alleged harvesting of publisher articles and YouTube videos for Gemini model training without consent or remuneration. The inquiry highlights the growing tension between AI model development and intellectual‑property rights in the digital economy.
Computer World
Aller aux sources
17 sources citées