Actualités — Page72

Security researchers demonstrated that AI chatbots offering web browsing can be manipulated to act as covert command‑and‑control conduits for malware. By prompting the chatbot to fetch a malicious URL and parsing the returned text for instructions, attackers can hide malicious traffic behind legitimate AI service requests. The technique works without needing developer APIs or API keys, and can also be used to exfiltrate data. Experts recommend treating AI web‑enabled services like any other high‑trust cloud application, monitoring for abnormal automation patterns, and restricting access to managed devices.Lire la suite

Freeform, a metal‑3D‑printing startup, announced a $67 million Series B round led by a group of venture firms. The capital will fund upgrades to its GoldenEye laser‑fusion system and the development of a next‑generation platform called Skyfall, which will employ hundreds of lasers and AI‑driven simulations. The company aims to expand its workforce and facility to meet a growing backlog of mission‑critical parts, leveraging on‑site Nvidia H200 GPU clusters for real‑time physics modeling.Lire la suite

Reliance Industries chairperson Mukesh Ambani unveiled a ₹10 trillion (about $110 billion) plan to build AI computing infrastructure across India over the next seven years. The initiative includes gigawatt‑scale data centers, a nationwide edge‑computing network, and AI services integrated with the Jio telecom platform. Powered by surplus green energy, the project aims to lower the cost of AI compute, partner with Indian enterprises and academia, and embed AI across sectors such as manufacturing, logistics, agriculture, healthcare, and finance.Lire la suite

OpenAI has teamed up with India’s Tata Group to secure an initial 100 megawatts of AI‑ready data center capacity, with plans to expand toward 1 gigawatt. The partnership, part of OpenAI’s Stargate project, will make OpenAI the first customer of Tata Consultancy Services’ HyperVault business and will roll out ChatGPT Enterprise across Tata’s workforce. The collaboration aims to reduce latency for Indian users, meet data‑residency requirements, and accelerate enterprise AI adoption throughout the country.Lire la suite

OpenAI has partnered with Indian fintech firm Pine Labs to embed its application programming interfaces into the company’s payments and commerce platform. The collaboration aims to automate settlement, reconciliation and invoicing workflows, extending AI‑driven efficiencies from Pine Labs’ internal operations to merchants and corporate clients. The move reflects OpenAI’s broader push to deepen its presence in India beyond consumer‑facing tools, while Pine Labs seeks to boost B2B adoption of AI agents for high‑volume financial tasks. The partnership is non‑exclusive, does not involve revenue sharing, and includes new security and compliance measures.Lire la suite

Sarvam AI, a Bengaluru‑based startup, says its Sarvam Vision model outperforms global rivals Gemini and ChatGPT on key optical character recognition (OCR) benchmarks for Indian languages. The model supports all 22 scheduled Indian languages and can handle complex tables, charts, and real‑world scene text. Paired with the Bulbul V3 text‑to‑speech system, which offers 35 local‑accented voices, the company positions itself as a builder of "sovereign AI" tailored to India’s linguistic diversity. Sarvam hopes its technology will help small businesses and government agencies digitize records more accurately and spur broader AI innovation focused on regional needs.Lire la suite

OpenAI has introduced ChatGPT Health, a dedicated health‑focused tab within the ChatGPT app that offers users a safer way to ask medical questions, review lab results, and organize health information. The feature uses the same large language model as standard ChatGPT but adds stricter limits, physician‑reviewed responses, and extra encryption to protect sensitive data. It can sync with apps such as Apple Health and upload documents, yet it does not replace professional diagnosis or treatment. OpenAI stresses that the tool is for consumer wellness and is not HIPAA‑covered, while acknowledging ongoing risks like hallucinations and the need for user caution.Lire la suite

OpenAI announced a partnership with six public and private higher‑education institutions in India, aiming to provide campus‑wide access to its ChatGPT Edu tools, faculty training, and responsible‑use frameworks. The initiative targets more than 100,000 students, faculty, and staff and includes collaborations with Indian ed‑tech platforms to offer structured AI courses. By embedding AI into core academic workflows such as coding, research, and analytics, OpenAI seeks to accelerate AI skill development and shape how artificial intelligence is taught and governed within one of the world’s largest higher‑education systems.Lire la suite

Generative AI is producing a flood of low‑quality, repetitive content—dubbed “AI slop”—that now dominates social‑media feeds and academic publishing. Creators such as Rosanna Pansino are responding by recreating AI‑generated videos with real‑world skill, while platforms, researchers, and regulators explore labeling, watermarking, and new policies to curb the spread. The battle pits human creativity against automated content machines, highlighting concerns about misinformation, deepfakes, and the future of authentic online experiences.Lire la suite

Mistral AI chief executive Arthur Mensch says companies are "replatforming," moving from traditional software to AI-driven solutions. He warns that success depends on having the "right infrastructure"—including clean data, cloud and compute resources, security, and skilled staff. Mensch predicts that more than half of current enterprise SaaS applications could be replaced by AI tools, creating a gap between firms that adopt AI and those that do not. He sees the trend as a major growth opportunity for Mistral, noting that over 100 enterprise customers are already exploring the shift.Lire la suite

Google's NotebookLM tool now lets users revise individual slides with natural‑language prompts and export decks as PowerPoint‑ready PPTX files. The update aims to streamline the slide‑creation workflow by allowing targeted edits without regenerating whole decks, while also preparing for future Google Slides export support. Users are cautioned that extensive revisions may affect layout consistency, requiring manual cleanup.Lire la suite

Perplexity, an AI search startup, is phasing out advertising and focusing on paid subscriptions for business users and high‑end professionals. Executives say ads could erode user trust, so the company will prioritize accuracy and revenue from customers like finance experts, lawyers, doctors, and CEOs. While not ruling out future ads, Perplexity aligns itself with the anti‑ad camp in the generative‑AI industry, contrasting with rivals such as OpenAI, which is testing ads, and Anthropic, which remains ad‑free.Lire la suite

Cameo, the platform that lets celebrities sell short personalized videos, secured a preliminary victory in its trademark lawsuit against OpenAI. A California judge ruled that OpenAI’s Sora video‑generation feature cannot use the term “Cameo” or any confusingly similar variation. The decision includes a preliminary injunction that halts the use of the name, marking another notable intellectual‑property clash as AI companies expand video‑creation capabilities.Lire la suite

Anthropic announced the release of Claude Sonnet 4.6, an upgraded mid‑range AI model that can code at a level comparable to its larger Opus series and interact with computers much like a human user. The model demonstrated human‑baseline performance on the OSWorld benchmark, handling tasks such as form filling and tab switching without specialized connectors. Anthropic also highlighted improved resistance to prompt‑injection attacks and a beta‑tested 1 million‑token context window, signaling stronger safety and scalability. The launch coincides with a surge in Claude’s popularity and a high‑profile advertising campaign targeting rival OpenAI.Lire la suite

A federal district court in Northern California ruled in favor of the video‑message platform Cameo, ordering OpenAI to cease using the word “Cameo” for its AI‑powered video generation feature. The court found the name likely to cause user confusion and rejected OpenAI’s claim that the term was merely descriptive. OpenAI subsequently renamed the feature “Characters.” The decision marks a significant win for Cameo’s brand protection efforts amid a series of recent intellectual‑property disputes involving OpenAI.Lire la suite

OpenAI announced a partnership with Peter Steinberger, the founder of the open-source AI assistant OpenClaw. Steinberger will join OpenAI to help expand personal AI agents while transitioning OpenClaw to an independent foundation that preserves its open-source roots. The deal provides OpenAI with credibility in the developer community and access to a viral platform known for autonomous task execution across messaging apps. Both parties view the collaboration as a catalyst for making personal AI agents a mainstream tool.Lire la suite

Infosys announced a partnership with Anthropic to integrate the latter's Claude models into its Topaz AI platform, creating autonomous AI agents for complex enterprise workflows in sectors such as banking, telecoms and manufacturing. The deal was unveiled at India's AI Impact Summit in New Delhi amid concerns that large‑language‑model tools could disrupt the Indian IT services industry. Infosys will use Anthropic's Claude Code for software development tasks and has already begun internal deployments. The collaboration also offers Anthropic a pathway into regulated enterprise markets, leveraging Infosys' industry expertise.Lire la suite

Fear of missing out on artificial intelligence—AI FOMO—is shaping how companies invest in technology and how employees view their jobs. Research shows that many leaders adopt AI out of anxiety rather than strategic need, while workers worry about skill relevance and autonomy. Higher AI literacy reduces the fear, but the pressure to keep pace creates a feedback loop of rushed adoption and mixed results. The trend underscores the need for intentional, purpose‑driven AI implementation rather than reactionary moves driven by fear.Lire la suite

Apple Music is rolling out two major upgrades in its latest iOS update: a new AI‑driven playlist creator called Playlist Playground and a Live Events hub for upcoming gigs. Playlist Playground, currently in beta, lets users generate 25‑track playlists from text prompts or a selection of at least ten songs, and it works on devices without Apple Intelligence. The Live Events feature will surface local concerts within the app, resembling Spotify’s similar offering. Both updates aim to broaden music discovery and personalize the listening experience.Lire la suite

Ireland's Data Protection Commission has opened a large‑scale inquiry into X (formerly Twitter) concerning the AI tool Grok, which allegedly produced millions of sexualized images—including thousands depicting children—without consent. The investigation will assess X’s compliance with GDPR obligations and follows a prior European Commission probe under the Digital Services Act. The Center for Countering Digital Hate’s recent review highlighted the scale of the issue, and X’s own statements about restricting Grok’s capabilities have been called into question by on‑the‑ground testing.Lire la suite