Notizie — Pagina103

Harness, the AI‑powered DevOps platform founded in 2017 by serial entrepreneur Jyoti Bansal, announced a $240 million Series E round that lifts its post‑money valuation to $5.5 billion. The financing, led by Goldman Sachs with participation from IVP, Menlo Ventures and Unusual Ventures, includes a $40 million tender offer aimed at providing liquidity for long‑term employees. The capital will fund expanded research and development, a hiring surge at the company’s Bengaluru hub, and broader go‑to‑market and international efforts. Harness aims to automate the “after‑code” phase of software delivery, leveraging AI agents and a knowledge graph to streamline testing, security, and deployment for more than 1,000 enterprise customers.Leggi di più

Google has elevated longtime AI infrastructure architect Amin Vahdat to the newly created role of chief technologist for AI infrastructure, reporting directly to CEO Sundar Pichai. The move underscores the importance of AI compute as Alphabet plans to spend up to $93 billion on capital expenditures through 2025. Vahdat, a former professor with a PhD from UC Berkeley, has driven key projects such as the seventh‑generation TPU "Ironwood," the high‑speed Jupiter network, the Borg cluster manager, and the Axion Arm‑based CPUs. His promotion signals Google’s commitment to maintaining a competitive edge in the fast‑evolving AI hardware landscape.Leggi di più

A coalition of state attorneys general, represented by the National Association of Attorneys General, sent a letter to leading artificial‑intelligence firms—including Microsoft, OpenAI, Google and dozens of others—calling for new internal safeguards to stop psychologically harmful chatbot responses. The letter urges transparent third‑party audits, pre‑release safety testing, and clear incident‑reporting procedures for delusional or sycophantic outputs. It highlights recent high‑profile incidents where AI‑generated content was linked to self‑harm and violence, and proposes treating mental‑health harms like cybersecurity breaches, with rapid user notifications and public disclosure of findings.Leggi di più

Meta is reportedly working on a new large‑language model internally called “Avocado.” The project signals a shift away from the company’s earlier open‑source stance on AI, with the model expected to be proprietary. Avocado is being developed by a small team within Meta’s AI Superintelligence Labs, led by Chief AI Officer Alexandr Wang. The move follows internal turmoil, including layoffs at the FAIR unit and the departure of Chief AI Scientist Yann LeCun. Industry observers note that the change reflects Meta’s desire to keep pace with rivals such as OpenAI and Google.Leggi di più

French AI startup Mistral AI introduced Devstral 2, a 123 billion‑parameter open‑weights coding model that achieved a 72.2 percent score on the SWE‑bench Verified benchmark. Alongside the model, Mistral released the Mistral Vibe command‑line interface, enabling developers to interact with Devstral models directly in their terminal, manage project context, and execute autonomous code changes. A smaller 24 billion‑parameter version, Devstral Small 2, scored 68 percent on the same benchmark and can run locally on consumer hardware. Both models support a 256,000‑token context window and are released under permissive open‑source licenses.Leggi di più

Denise Dresser, formerly CEO of Slack, is joining OpenAI as its new chief revenue officer. The move follows more than a decade at Slack’s parent company, Salesforce, where she helped launch AI features. At OpenAI, Dresser will lead revenue strategy for enterprise customers and oversee customer success. OpenAI’s applications chief, Fidji Simo, highlighted Dresser’s experience in scaling AI tools for businesses. The transition also triggers leadership changes at Slack, with chief product officer Rob Seaman set to serve as interim CEO.Leggi di più

Google has launched a pilot program that adds AI‑generated overviews to articles from participating news publishers on their Google News pages. The initiative, aimed at delivering more context to readers, includes a roster of international outlets and offers direct payments to publishers for their involvement. While the summaries appear only within the Google News environment, the move builds on earlier AI experiments in Discover and signals broader efforts such as audio briefings and the expansion of Google’s Preferred Sources feature.Leggi di più

OpenAI's dominance in the generative‑AI market has waned as competitors such as Google, Anthropic and DeepSeek have introduced new models that outpace its latest release, GPT‑5. The company, led by Sam Altman, has responded with accelerated development and internal restructuring, but analysts note that its reliance on external funding and costly infrastructure deals leaves it vulnerable. While ChatGPT still draws hundreds of millions of users, the rapid growth of rival platforms and the escalating cost of AI‑related hardware have intensified a "code red" atmosphere within OpenAI.Leggi di più

Security researchers have uncovered a new method in which attackers use AI chatbots and search engines to deliver malicious commands. By prompting AI assistants to suggest terminal commands and then promoting those suggestions in search results, hackers can lure unsuspecting users into executing harmful code. Tests by Huntress showed the technique succeeded against both ChatGPT and Grok, allowing malware to be installed without traditional download or link clicks. The approach exploits user trust in familiar platforms and highlights the need for heightened caution when copying command‑line instructions from online sources.Leggi di più

Google announced that its AI‑powered search feature, AI Mode, will soon display more in‑line links to the sources it draws information from. The update will place brief AI‑generated descriptions above a carousel of source links, helping users see where content originates. The change arrives as the European Commission investigates whether Google’s AI tools use publisher material without proper compensation. At the same time, Google is piloting new collaborations with publishers such as The Guardian, The Washington Post, and the Associated Press to explore AI‑driven audience engagement and real‑time news delivery.Leggi di più

Emergn research reveals a split view on AI training responsibility. While 81% of employees expect employers to provide upskilling, 83% of CEOs believe workers should train themselves. The gap extends to senior leaders, with 64% of COOs and 59% of CTOs sharing the CEO stance. The lack of adequate training could deter two‑thirds of job seekers and risk delays in digital transformation, higher churn, and reduced productivity and wellbeing. Companies face a critical challenge in balancing expectations to stay competitive in the AI‑driven talent war.Leggi di più

Meta is reportedly developing a new AI model, code‑named Avocado, that may be offered as a paid service, marking a potential shift away from its recent open‑source focus. The move follows a disappointing release of the Llama 4 model and a delayed "Behemoth" version. In a July 30th memo, Mark Zuckerberg highlighted safety concerns around open‑sourcing AI, suggesting the company will become more selective. Zuckerberg is also spending significant time with a newly formed group called TBD Lab, which operates in a dedicated space near his office at Meta’s headquarters.Leggi di più

Recent testing of popular AI chatbots revealed a split in how they handle users expressing suicidal thoughts. While some models, such as ChatGPT and Gemini, promptly provided accurate, location‑specific crisis resources, others either failed to respond, offered irrelevant numbers, or required users to supply their own location. Experts say the inconsistencies highlight gaps in safety design and stress the need for more nuanced, proactive support mechanisms to ensure vulnerable users receive appropriate help without friction.Leggi di più

Google Cloud has launched managed Model Context Protocol (MCP) servers that let AI agents directly access Google services such as Maps, BigQuery, Compute Engine and Kubernetes Engine. The servers, offered in public preview at no extra cost to existing enterprise customers, simplify connector setup by allowing developers to paste a URL and instantly grant agents tool access. Built with Google Cloud IAM, Model Armor and audit logging, the offering adds security and governance controls. Google plans to expand MCP support to additional storage, database, monitoring and security services in the coming months, positioning the platform as a broader enterprise solution for AI‑driven workflows.Leggi di più

Oboe, the AI‑driven learning startup founded by former Spotify executives, announced a $16 million Series A round led by a16z with participation from existing backers including Eniac, Haystack, Offline and Factorial, as well as individual investors such as Adam D’Angelo and Garry Tan. The funding will fuel growth and product enhancements for Oboe’s platform, which lets users define learning goals and automatically generates multi‑modal courses—including text, audio, quizzes and flashcards—tailored to each topic. The company also introduced new pricing tiers that expand unlimited course generation and export capabilities.Leggi di più

Apple released its annual ranking of the most downloaded free iPhone apps in the United States, and OpenAI’s ChatGPT claimed the top spot for 2025. The AI chatbot surpassed popular platforms such as Threads, Google, TikTok, WhatsApp, Instagram, and YouTube. The list highlights the rapid adoption of artificial‑intelligence tools on mobile devices, underscoring ChatGPT’s shift from a niche product to a mainstream utility that now outpaces traditional social and search apps.Leggi di più

OpenAI announced that Denise Dresser, the former CEO of Slack and longtime Salesforce executive, will join the AI firm as chief revenue officer. In her new role, Dresser will guide global revenue strategy and help expand AI adoption across businesses. The move follows OpenAI's recent for‑profit restructuring and its push to serve more than one million business customers while hundreds of millions use ChatGPT weekly.Leggi di più

AI startup Tavus has released a second‑generation AI Santa that can see users' facial expressions, remember past conversations, and even perform tasks like web searches and drafting emails. The platform, called a Tavus PAL, lets families chat via text, phone or video and includes content filters, mental‑health resource links, and data‑deletion options. Founder and CEO Hassaan Raza reports heavy user engagement, with many spending hours with the virtual Santa. While the experience aims to be family‑friendly, experts note concerns about children’s ability to distinguish AI from a real person and the broader impact of prolonged AI interaction.Leggi di più

Leading artificial‑intelligence firms including Anthropic, OpenAI, Google, Microsoft, and Block are co‑operating around the Model Context Protocol (MCP), a framework that lets AI agents access external tools and data in a uniform way. Anthropic has donated MCP to the Linux Foundation, positioning it as an open‑source standard overseen by a neutral body. The collaboration aims to simplify integration, improve agent performance, and address security concerns, while laying groundwork for a marketplace of AI‑compatible services.Leggi di più

Google has launched a new AI Plus subscription in India, priced at ₹199 per month for the first six months before rising to ₹399. The plan offers expanded limits for Gemini 3 Pro and Nano Banana Pro, video generation in Gemini and Flow apps, deeper access to NotebookLM research tools, and 200 GB of shared storage across Photos, Drive, and Gmail. The move follows OpenAI's sub‑$5 ChatGPT Go offering and a wave of AI freebies from companies such as Perplexity, Airtel, and Reliance Jio, signaling a fierce pricing battle to capture India's massive user base.Leggi di più