Actualités — Page56

A team of researchers at Purdue University has created a privacy‑focused technique that lets users edit photos with AI while keeping sensitive facial data on the device. The method masks designated regions, such as faces, before the image is sent to an AI service, uploads only the non‑masked portion, and then seamlessly reintegrates the original masked area after editing. The approach works with existing commercial generative AI models, requires no model retraining, and has been validated by testing AI classifiers on masked versus unmasked images, showing a dramatic drop in attribute‑recognition accuracy. The researchers have published their findings in IEEE Transactions on Artificial Intelligence and filed a patent, positioning the technology for future commercial adoption.Lire la suite

Google’s senior vice president of knowledge and information, Nick Fox, told WIRED that the company is not ruling out advertising in its Gemini AI product. While the firm is testing ads in AI Mode, a Search‑powered experience, it says any future ads would be clearly labeled and relevant to users. Fox emphasized that Google’s strong financial position allows it to prioritize product quality over immediate monetization, and that lessons learned from AI Mode could inform any later Gemini ad strategy. The discussion also touched on the new Personal Intelligence feature and its potential impact on ad targeting.Lire la suite

Claude has rolled out a beta feature that embeds interactive charts, diagrams and other visual tools directly into chat conversations. The new capability lets users adjust controls and instantly see graphics update, turning static explanations into dynamic, on‑the‑fly demonstrations. Early tests show practical applications such as coffee‑to‑water calculators, volcanic cross‑sections and even fashion‑style generators, highlighting the potential for more engaging AI‑driven interactions.Lire la suite

Rox, a sales‑automation startup that builds autonomous AI agents to boost revenue productivity, has closed a new funding round that values the company at $1.2 billion. The round was led by returning backer General Catalyst and follows earlier seed and Series A investments that together total $50 million. Founded in 2024 by former New Relic chief growth officer Ishan Mukherjee, Rox positions its platform as an intelligent revenue operating system that plugs into existing software stacks and automates tasks such as account monitoring, prospect research, and CRM updates. Customers include Ramp, MongoDB, and New Relic, and the company competes with established revenue‑intelligence firms and emerging AI‑native CRM solutions.Lire la suite

Meta has announced the purchase of Moltbook, a platform built primarily for artificial intelligence agents to communicate with one another. The network currently hosts AI bots that share updates, code snippets, and philosophical reflections while human users watch from the sidelines. The acquisition fits Meta's broader push into AI‑driven digital assistants and could serve as a testbed for how autonomous agents collaborate online. Industry observers note that the move raises questions about the authenticity of social feeds, as bots can generate endless content and blur the line between human and machine interaction.Lire la suite

Google and Samsung have introduced a new Gemini feature that lets users automate tasks in apps through simple prompts. Starting with food delivery and rideshare services, the assistant can navigate app interfaces in a virtual window, fill in details, and pause before final confirmation, giving users control over each step. Early testing shows the system can handle requests like ordering a ride to the airport or a coffee and croissant, asking clarifying questions when needed. The rollout marks a significant step forward for AI assistants on mobile devices.Lire la suite

Anthropic has introduced a new beta feature that lets Claude generate interactive visualizations—HTML and SVG diagrams—directly within chat. The tool aims to replace dense text with clear, on‑demand graphics, helping users grasp complex steps such as changing a tire or exploring a periodic table. Available on web and desktop for all plan types, the feature automatically creates clickable visuals when appropriate or when users request them, offering a more dynamic conversational experience.Lire la suite

Anthropic has updated its Claude chatbot to create custom charts, diagrams, and other visualizations within the chat window. When the system judges a visual to be useful, it inserts the image inline, allowing users to see data and concepts without leaving the conversation. Users can also request visuals directly. The new feature rolls out to all users by default and complements Anthropic’s existing "artifacts" side‑panel tools, which remain persistent. Similar visual capabilities have recently appeared in OpenAI’s ChatGPT and Google Gemini, marking a broader shift toward interactive AI‑generated graphics.Lire la suite

Anthropic announced that its Claude chatbot can now produce charts, diagrams, and other visual aids directly within its responses. The feature, built on HTML and XML vector graphics, functions like a digital whiteboard and is available to all users, free or paid, though it is still in beta and not yet supported on mobile devices. Anthropic emphasizes that the update is not an image‑generation tool and may exhibit some quirks as it rolls out.Lire la suite

Physical AI embeds artificial intelligence in machines that can sense their surroundings, make real‑time decisions, and act in the physical world. From autonomous vehicles and warehouse robots to surgical assistants and smart‑city systems, these technologies blend sensors, computer vision, machine learning, and reinforcement learning to close the perception‑decision‑action loop. While early deployments already exist, challenges around safety, reliability, edge‑case handling, and costly real‑world training remain central as the field pushes toward broader, embodied AI applications.Lire la suite

Google Maps is adding a Gemini‑driven conversational feature called Ask Maps that lets users pose natural‑language questions about routes, places and real‑world needs. At the same time, the app receives an Immersive Navigation update with a 3D view, detailed road information and more natural voice guidance. Both upgrades aim to personalize travel planning and make driving safer and more intuitive, rolling out initially on Android and iOS in the United States and India.Lire la suite

An Anthropic guide shows how specific prompting techniques—playful constraints, role‑playing, future‑historian perspectives, imagined debates, and structured analyses—can coax Claude into delivering clearer, more imaginative, and better‑structured answers. By framing queries with creative frames, users can tap Claude’s strength in analogies, layered explanations, and multi‑view reasoning, turning ordinary questions into insightful, engaging conversations.Lire la suite

Google is rolling out a new Gemini AI feature called Ask Maps to the Google Maps app. The conversational experience lets users ask follow‑up questions, rephrase queries and explore prompts about trips, nearby amenities and tourist spots. Ask Maps draws on users’ past Maps data to tailor recommendations, such as locating charging stations or well‑lit tennis courts. The update also makes spoken directions sound more natural and improves route alternatives and entrance details. The feature is live for users in the United States and India and will later expand to CarPlay and Android Auto.Lire la suite

Perplexity announced Personal Computer, software that turns a Mac mini into an always‑on AI agent that links local files, apps and cloud services. The product is limited to Perplexity Max subscribers at $200 per month with 10,000 compute credits and includes audit trails and a kill switch. At the same event the company unveiled an enterprise version with SOC 2 Type II compliance, SAML single sign‑on, audit logs and sandboxing, and native integrations with Snowflake, Salesforce, HubSpot and other platforms. Finance data tools were expanded, and new developer APIs were introduced, underscoring Perplexity’s push to sell AI orchestration rather than a single model.Lire la suite

Anthropic announced new updates that tightly link its Claude AI assistant with Microsoft Excel and PowerPoint. The enhancements let users maintain a single, continuous conversation across both applications, eliminating repetitive copy‑and‑paste. Claude can read spreadsheet data and insert it directly into presentations, while reusable "Skills" let organizations save and share automated workflows with a single click. A preloaded set of starter Skills for common tasks is also included. The features are available to paid‑plan users on both Mac and Windows platforms.Lire la suite

The State Department has removed Anthropic’s Claude model from its internal chatbot after a directive from President Trump. The Treasury and Health & Human Services departments are also ending Claude use, urging staff to adopt ChatGPT or Google’s Gemini. Meanwhile, the Senate has approved the use of Gemini, ChatGPT, and Microsoft Copilot for official work, outlining specific tasks for the tools. The changes reflect a broader realignment of AI policy across federal agencies.Lire la suite

Breakout Ventures has closed a $114 million Fund III aimed at early‑stage startups that apply artificial intelligence to scientific fields such as biology and chemistry. The firm, which spun out of a Thiel Foundation grant program, has already written checks to three companies and plans to invest in at least 20 companies, with check sizes ranging from $500,000 to $5 million. Limited partners include The Kraft Group, Pinegrove Venture Partners and S‑Cubed Capital. Managing Director Lindy Fishburne emphasized the firm’s focus on founders who can unlock complex scientific problems with AI.Lire la suite

Replit announced a $400 million Series D funding round that lifts its valuation to $9 billion. The round was led by Georgian Partners and included investors such as G Squared, Prysm Capital, Coatue, Andreessen Horowitz, Craft Ventures, Y Combinator, Accenture Ventures, Okta Ventures, and Databricks Ventures. Founder and CEO Amjad Masad noted additional backing from angel investors Shaquille O’Neal and Jared Leto. The new capital follows a previous $250 million raise that pushed the company’s valuation to $3 billion and placed its annualized revenue on track for $150 million, with a goal of $1 billion in annual recurring revenue by year‑end.Lire la suite

Grammarly announced that it is disabling its Expert Review AI agent after receiving critical feedback from experts who said the tool misrepresented their voices. CEO Shishir Mehrotra explained that the company will rethink the feature to give experts real control over how, or if, their knowledge is used. The move reflects Grammarly’s commitment to responsible AI use and opens the platform for experts who want to build their own agents while maintaining full authority over representation.Lire la suite

Nvidia announced a $26 billion investment over the next five years to create open-weight artificial‑intelligence models, marking a shift from pure chip manufacturing to a broader AI research role. The company unveiled Nemotron 3 Super, its most capable open‑weight model to date, featuring 128 billion parameters and claiming top performance on several benchmarks. Executives highlighted the strategic aim of fostering an ecosystem that leverages Nvidia’s hardware while offering publicly available model weights for startups and researchers. Industry observers see the move as a significant signal of Nvidia’s commitment to openness and a potential counterbalance to Chinese open‑source AI efforts.Lire la suite