Revue de presse générée automatiquement avec
Profil généré automatiquement
Pulse
Un individu à l'aise avec la technologie, passionné par l'intersection du logiciel open source, de l'intelligence artificielle et des télécommunications, avec un intérêt vif pour les modèles et les tendances émergents qui façonnent le paysage technologique. Il suit les développements et les innovations du secteur, à la recherche d'informations pour éclairer son travail.
Open source (20%)Models (20%)Ai (20%)Telecomms (20%)Technology trends (20%)
Vous souhaitez recevoir chaque jour la revue de presse de ce profil ?
AI code quality, Generative video, AI ecosystem, Tech trends...
Jeudi 18 décembre 2025 à 15:50
AI Insights
AI‑generated code still riddled with bugs
Tech Radar warns that code produced by generative AI tools contains more defects than human‑written equivalents, even though it reduces typographical errors. The report underscores the continued need for human code reviewers to catch logical flaws and maintain software reliability.
Tech Radar
Anthropic’s automated account shutdown sparks governance concerns
Computer World details how Anthropic’s AI‑driven policy enforcement abruptly disabled a Swiss cybersecurity firm’s account, erasing 80 % of its projects until a paid reinstatement restored them. The incident highlights the emerging risk of vendor‑controlled “kill switches” and the lack of transparent appeal mechanisms for enterprise users.
Computer World
One‑third of UK citizens turn to AI for emotional support
The Guardian cites an AI Security Institute survey revealing that 33 % of Britons have used chatbots such as ChatGPT or Alexa for companionship, with 10 % doing so weekly. This growing reliance raises questions about the mental‑health implications of AI‑mediated social interaction.
The Guardian
AI’s carbon and water footprints reach alarming levels
A joint investigation by The Guardian and The Verge finds that the AI boom in 2025 emitted as much CO₂ as New York City and consumed more water than the global bottled‑water market. Researchers blame opaque data‑center reporting and the surge in large‑scale model training for the environmental strain.
The Guardian
The Verge
Emerging Models
Luma AI launches start‑to‑end frame video generation
TechCrunch reports that Luma’s new model, available on its Dream Machine platform, can synthesize full‑length video clips from just a start and end frame, pushing generative video capabilities toward mainstream adoption. The technology promises to streamline content creation for marketers and designers.
TechCrunch
Apple’s SHARP model creates 3D scenes from a single photo in under a second
Wccftech and Tech Radar both highlight Apple’s SHARP AI, which converts a 2D image into a photorealistic 3D environment in less than a second, opening new possibilities for VR memory capture and rapid prototyping. Apple released a detailed study demonstrating the model’s speed and accuracy.
Wccftech
Tech Radar
Google accelerates Gemini with the Flash variant
The Verge and Computer World note that Google’s Gemini 3 Flash delivers the reasoning power of Gemini 3 Pro with reduced latency and lower compute cost, making it the default model in the Gemini app and available via Vertex AI. Early benchmarks show performance on par with GPT‑5.2 on standard tests.
The Verge
Computer World
Meta’s LeCun targets a €3 bn valuation for new AI start‑up
The Financial Times reveals that Yann LeCun is in talks to raise €500 million for a nascent AI venture, aiming for a €3 billion valuation ahead of its January launch. The move signals Meta’s intent to spin out cutting‑edge research into a market‑ready entity.
Financial Times
Open Source & Ecosystem
OpenAI unveils an app store inside ChatGPT, backed by open‑source SDKs
Engadget covers OpenAI’s new app directory that lets developers publish “apps” directly within ChatGPT, supported by an open‑source UI library and SDK. The platform enables integrations with services like Dropbox and Spotify, expanding the conversational AI ecosystem.
Engadget
Adobe faces a class‑action lawsuit over alleged misuse of copyrighted works in AI training
TechCrunch reports that a proposed class‑action accuses Adobe of incorporating artists’ copyrighted material into its generative AI models without permission, reflecting broader legal pressures on firms that rely on large, uncurated datasets. The case could set precedents for data‑rights compliance in AI development.
TechCrunch
Technology Trends & Industry Moves
Coursera‑Udemy merger accelerates AI‑driven workforce upskilling
The Information and Computer World explain that the $950 million stock deal will combine Coursera’s academic reach with Udemy’s enterprise‑focused AI reskilling platform, creating a unified system to deliver bite‑sized, AI‑aligned learning at scale. Executives stress the need to keep curricula in sync with rapid AI advancements.
The Information
Computer World
DRAM price surge blamed on AI‑driven demand
PC Gamer and Wccftech report that memory maker G.Skill attributes the 3‑4× rise in DDR5 prices to soaring AI workloads and constrained supply, warning buyers to assess cost‑benefit before upgrading. The surge underscores the broader hardware bottlenecks fueling AI expansion.
PC Gamer
Wccftech
UK actors and performers vote to block digital scanning for AI use
The Guardian and Variety detail a landslide vote by the Equity union to refuse on‑set body scans, aiming to protect performers’ likenesses from being harvested for generative AI. The move marks a significant collective‑action stance against unchecked AI exploitation in the entertainment sector.
The Guardian
Variety
Aller aux sources
19 sources citées