Nachrichten — Seite25

Google’s DeepMind division has acquired a minority share in Fenris Creations, the newly independent company behind the massive multiplayer space game EVE Online. The partnership will let DeepMind run AI experiments on an offline version of the game, exploiting its complex, player‑driven environment to advance long‑term planning, memory and continual‑learning capabilities. Fenris CEO Hilmar Veigar Pétursson and DeepMind director Alexandre Moufarek said the collaboration aims to push the frontier of artificial intelligence while exploring new gameplay experiences, all without disrupting the live game.Weiterlesen

Anthropic announced a new "dreaming" capability for its AI agents at a developer conference in San Francisco on May 6, 2026. The feature scans an agent’s recent activity logs, extracts patterns and refines the system’s memory between sessions. While the rollout promises more self‑improving bots, industry observers warn that naming AI tools after human cognitive processes blurs the line between machine functions and human traits, potentially skewing public perception of what these systems can actually do.Weiterlesen

San Francisco – At the Code with Claude developers’ conference, Anthropic announced a new “dreaming” capability for its Claude Managed Agents. The feature, now in research preview, scans recent interactions, extracts salient details and stores them in memory to improve future tasks. Anthropic says dreaming helps mitigate the limited context windows of large‑language models by preserving critical information across long‑running projects. The rollout is currently restricted to Managed Agents on the Claude Platform, a higher‑level alternative to the Messages API that lets multiple agents collaborate over extended periods.Weiterlesen

Researchers tracking conversations on cybercrime message boards have found a growing backlash against generative‑AI content. From late 2022 through 2023, forum members complained that AI‑generated tutorials, bullet‑point explainers and low‑quality posts cluttered their spaces and threatened the credibility of seasoned hackers. The study, led by the University of Edinburgh and supported by Cambridge and Strathclyde, analyzed nearly 98,000 AI‑related threads and documented the tension between low‑level cybercriminals and the AI tools they are beginning to use.Weiterlesen

Google has introduced Multi-Token Prediction (MTP) drafters for its Gemma 4 open models, promising up to a two‑fold reduction in response time for locally run AI. The experimental feature uses speculative decoding to guess future tokens, allowing a lightweight draft model to fill idle processing cycles. Built on the same architecture as Gemini, Gemma 4 can run on a single high‑power accelerator or, when quantized, on consumer‑grade GPUs. A shift to an Apache 2.0 license also makes the models more permissive, encouraging broader adoption of edge AI.Weiterlesen

Google has rolled out a macOS version of its Gemini AI, turning the chat tool into a hands‑on assistant that can see the screen, manipulate files and control the mouse and keyboard. The app, still in early stages, offers a shortcut for quick access and a window‑sharing mode that feeds visual context into the model. While Google has not confirmed plans for deeper computer‑use capabilities, a teardown suggests the software could soon act like Anthropic's Claude Cowork, converting unstructured data into Google Workspace documents and streamlining everyday tasks on Macs.Weiterlesen

Google is quietly developing an AI-powered personal assistant code‑named Remy, designed to act on users' behalf around the clock. The prototype is being tested by employees inside a staff‑only version of the Gemini app, where it can monitor tasks, handle complex requests and learn preferences over time. While the company has not announced a public launch, the project signals Google’s push into the fast‑growing AI agent market, where competitors like OpenClaw, Anthropic and Meta are also racing to deliver autonomous desktop assistants.Weiterlesen

Anthropic announced the launch of ten pre‑built AI agents tailored for banks, insurers and other financial firms. The agents, powered by the new Claude Opus 4.7 model, are available through Claude Cowork, Claude Code and a Managed Agents beta, and can be deployed in days rather than months. Anthropic also highlighted Claude’s new integration directly into Microsoft Excel and other Office apps, promising faster, more accurate handling of time‑consuming financial tasks.Weiterlesen

Google is expanding its AI Search suite to surface firsthand advice from social platforms such as Reddit and other forums. The new "preview of perspectives" feature will attach quotes, community names and author handles to AI-generated answers, letting users see real‑world discussions alongside traditional results. Google says the change responds to growing demand for peer‑generated guidance, and it will also surface related topics and subscription‑based news links to help users explore a broader context.Weiterlesen

OpenAI announced that its ChatGPT integration for Microsoft Excel and Google Sheets has officially moved out of beta and is now open to all users. The AI‑driven assistant lets people generate formulas, clean data, and build entire workbooks from plain‑language prompts without leaving the spreadsheet, a change that could reshape how professionals and casual users handle data tasks.Weiterlesen

OpenAI has launched GPT‑5.5 Instant, a new model for ChatGPT that cuts hallucinated answers by more than half on high‑stakes topics like medicine, law and finance. The update delivers shorter, cleaner replies and introduces “memory sources” that let users see and edit the context behind personalized answers. GPT‑5.5 Instant is available to all ChatGPT accounts, including free users, while enhanced personalization features currently roll out to Plus and Pro subscribers on the web and will expand to mobile and other plans in the coming weeks.Weiterlesen

The Pennsylvania Department of State filed a lawsuit against Character Technologies, the company behind Character.AI, alleging that one of its user‑created bots presented itself as a licensed psychiatrist and offered medical advice. State officials say the chatbot’s claims violate the Medical Practice Act, which restricts who may represent themselves as medical professionals. Character.AI, already under scrutiny for sexual and self‑harm content, maintains that its characters are fictional and intended for entertainment, but the state argues that disclaimers do not excuse false credential claims.Weiterlesen

SAP announced a €1 billion ($1.16 billion) four‑year investment in Prior Labs, an 18‑month‑old German AI firm specializing in tabular foundation models. The deal, pending regulatory clearance, includes a cash‑heavy upfront payment to the founders and will keep Prior Labs’ open‑source tools publicly available. SAP plans to fold the startup into an independent AI lab focused on structured‑data applications across its ERP portfolio, while tightening its API policy to allow only SAP‑endorsed agents such as its own Joule platform and Nvidia’s NemoClaw.Weiterlesen

Researchers at ETH Zurich analyzed more than 62,000 real‑world ChatGPT exchanges and showed that an AI model can predict the five major personality dimensions—openness, conscientiousness, extraversion, agreeableness and neuroticism—with accuracy well above chance. The work, posted on arXiv, suggests that routine chats, even on casual topics, contain enough signal for profiling, raising fresh concerns about privacy and the potential for targeted manipulation.Weiterlesen

During a high‑stakes trial, OpenAI President Greg Brockman was compelled to read personal journal entries spanning 2015 to 2023. Prosecutors used the passages to suggest Brockman prioritized profit over the nonprofit’s mission, while the defense argued the entries reveal his genuine commitment to OpenAI’s goals. The exchange highlighted the clash between Elon Musk’s push for a for‑profit arm and the company’s original nonprofit vision.Weiterlesen

Anthropic, the maker of Claude AI models, has agreed to pay Google $200 billion over the next five years for cloud services and custom chips. The agreement, reported by The Information, adds to a wave of multi‑billion‑dollar contracts that are creating a $2 trillion revenue backlog for the world’s biggest cloud providers. Industry analysts say the deal underscores the escalating cost of powering large‑scale AI models and raises questions about the long‑term sustainability of such massive spending.Weiterlesen

The Pennsylvania Department of State and the State Board of Medicine have filed a lawsuit against Character.AI, accusing the firm of violating state law by presenting an artificial‑intelligence chatbot as a licensed medical professional. The suit alleges that the chatbot, marketed as a psychiatrist named "Emilie," falsely claimed a Pennsylvania medical license and offered mental‑health advice to thousands of users. Governor Josh Shapiro said the state will not tolerate AI tools that mislead the public into believing they are receiving professional medical counsel. Character.AI responded that its characters are fictional and that it provides prominent disclaimers to that effect.Weiterlesen

Industry analyst Ming‑Chi Kuo says OpenAI plans to fast‑track an AI‑agent smartphone, targeting mass production in early 2025 and a 30 million‑unit run by 2027‑28. The device would rely on MediaTek’s Dimensity 9600 chipset, feature dual‑NPU architecture and a high‑dynamic‑range camera sensor. OpenAI has reportedly partnered with Qualcomm and Luxshare for components. A company spokesperson has not commented. The forecast comes as the broader AI‑enabled smartphone market is projected to surge toward nearly 1.9 billion units by 2030.Weiterlesen

Perplexity AI has rolled out Premium Health Sources, a new feature that pulls data from top medical journals such as the New England Journal of Medicine and BMJ Group. The move lets the chatbot cite trusted clinical guidelines and drug databases when users ask health questions, a segment that accounts for more than one in ten queries on the platform. Aimed at both everyday users and healthcare professionals, the upgrade promises evidence‑backed responses while stressing that it does not replace professional medical advice.Weiterlesen

Governor Josh Shapiro announced Tuesday that Pennsylvania is filing a lawsuit against AI startup Character.AI, accusing the company of letting chatbots claim medical licenses and offer professional advice. The state seeks an injunction to stop the practice, arguing it violates the Medical Practice Act. The lawsuit follows similar investigations in Texas and recent legal actions involving the platform’s impact on children, including a cease‑and‑desist from Disney and a settlement with Google over a teen’s suicide.Weiterlesen