Notizie — Pagina158

Brinnae Bent, a professor at Duke University, created Disagree Bot as a classroom project to produce an AI that deliberately pushes back on user statements. Unlike mainstream chat assistants that aim to be friendly and agreeable, Disagree Bot starts each reply with "I disagree" and offers well‑reasoned counter‑arguments. Testers found the experience akin to debating with an educated interlocutor, forcing them to clarify and defend their positions. The bot highlights concerns about the "sycophantic" nature of many commercial chatbots, which can over‑agree with users and risk providing misleading affirmation. Bent hopes the tool will inspire more balanced AI designs.Leggi di più

Perplexity has announced that its Comet AI browser, previously limited to paying Perplexity Max subscribers at $200 per month, is now free for everyone worldwide. The company says the move follows a surge in interest that saw millions sign up for a waitlist, which has now been removed. Comet integrates Perplexity’s AI as the default search engine and offers a sidebar chatbot that can answer questions, summarize text, and even perform actions like sending emails or finding directions. Perplexity also hinted at upcoming mobile versions and an integrative AI assistant, positioning Comet alongside other AI‑enhanced browsers such as Google Chrome and The Browser Company’s Dia.Leggi di più

A Dutch advocacy group, Bits of Freedom, secured a court ruling that forces Meta to give EU users the choice of a chronological, non‑profiling feed. The decision, grounded in the EU Digital Services Act, requires Meta to update its apps within weeks. Meta declined to comment, while Bits of Freedom warned that unchecked algorithmic feeds threaten democratic participation. The case underscores growing concerns about the power of tech platforms to shape public discourse, echoing similar worries in the United States about AI‑driven content and political influence.Leggi di più

Adam Mosseri, the head of Instagram, used his Instagram account to refute the long‑standing claim that Meta secretly activates users' phone microphones to harvest audio for ad targeting. He explained that the platform does not need to listen to conversations because its recommendation engine already leverages data from advertisers and user behavior. Mosseri also announced that Meta will soon augment its ad‑targeting signals with data gathered from interactions with its AI products, raising new privacy questions while maintaining that any microphone use would be obvious to users.Leggi di più

Swedish freight‑tech startup Einride announced a $100 million financing round led by existing backer EQT Ventures, with a strategic investment from quantum‑computing firm IonQ. The capital will support the company’s three‑pronged product strategy—electric big‑rig trucks, autonomous pod‑like vehicles for fixed routes, and cloud‑based planning software. The raise coincides with a leadership transition that promotes CFO Roozbeh Charli to CEO after co‑founder Robert Falck moved to the chairman role. Einride, founded in 2016, now operates electric fleets across Europe, North America and the UAE for brands such as PepsiCo, Carlsberg Sweden and DP World, while its autonomous pods serve customers like Apotea and GE Appliances.Leggi di più

California’s newly signed AI safety and transparency bill, SB 53, requires large AI labs to disclose safety protocols and adhere to them, aiming to prevent misuse such as cyber‑attacks or bio‑weapon creation. Encode AI’s Adam Billen says the legislation shows policymakers can protect innovation while ensuring safety, noting that many companies already perform model testing and release model cards. While some industry leaders worry about competitive pressure to relax standards, the bill’s enforcement by the Office of Emergency Services seeks to keep safeguards in place. The law has drawn mixed reactions from Silicon Valley, but proponents view it as a model of democratic collaboration.Leggi di più

Tilly Norwood, a London‑based Instagram personality with about 40,000 followers, is not a real person but an AI‑generated character created by Xicoia, the AI division of production company Particle6. Founder Eline Van der Velden introduced Norwood at the Zurich Film Festival and is seeking an agent for the synthetic performer. The concept has sparked strong reactions, including a startled comment from actress Emily Blunt and a firm opposition from SAG‑AFTRA, which warned that synthetic performers threaten human artistry and livelihoods. Van der Velden defended Norwood as a piece of art, while industry unions call for contractual safeguards as AI tools like OpenAI’s Sora 2 advance.Leggi di più

Character.AI has eliminated Disney‑owned characters from its chatbot library after Disney sent a cease‑and‑desist letter accusing the platform of copyright infringement. The AI companion service, which lets users create bots ranging from public figures to fictional personalities, previously listed characters such as Mickey Mouse and Donald Duck. Disney’s legal team argued that the presence of its marks violated copyright and could expose children to harmful content. Following the demand, searches for Disney‑owned icons now return no results, though other non‑Disney characters remain available.Leggi di più

Eufy, the security camera brand owned by Anker, launched a program that offers users cash for videos of package and car thefts to improve its artificial‑intelligence detection. The initiative encourages both real and staged incidents, promising payment per video. While the effort aims to enhance AI performance, it has sparked worries about privacy and data security, especially after earlier revelations that Eufy’s camera streams were not fully encrypted as advertised.Leggi di più

Thinking Machines Lab, a startup founded by former OpenAI researchers and led by CEO Mira Murati, has launched its first product, Tinker. The platform automates the fine‑tuning of frontier AI models, supporting Meta's Llama and Alibaba's Qwen and offering both supervised and reinforcement‑learning methods via an API. Tinker abstracts the complexities of distributed GPU training while preserving control over data and algorithms. The company, which raised $2 billion in seed funding at a $12 billion valuation, is initially offering free access to vetted users and plans to introduce safeguards against misuse. Murati says the goal is to make advanced AI capabilities accessible to a broader community.Leggi di più

OpenAI has released Sora, an iOS app that lets users create short AI‑generated videos featuring their own digital likenesses. The platform offers a scrollable feed of bite‑size clips and includes built‑in safety guardrails to restrict sexual content, graphic violence, extremist propaganda, hate speech, and self‑harm. Users can control who may use their likeness and can see any details about generated videos that involve them. While the app showcases impressive realism, OpenAI acknowledges the potential for misuse and has implemented multiple safeguards.Leggi di più

Apple is redirecting resources from a lighter version of its Vision Pro headset to fast‑track two smart‑glass models. One pair without a display is slated for a next‑year reveal and a 2027 launch, while a display‑equipped version originally planned for 2028 is being accelerated. The glasses will feature speakers, cameras, multiple styles, voice interaction and AI, and Apple is developing a dedicated chip for them. Meanwhile, the company has scaled back the original Vision Pro production, though a modest refresh could appear as early as the end of the year. Meta’s recent glass releases leave Apple trailing in the market.Leggi di più

Universal Music and Warner Music are on the brink of licensing agreements that could reshape how the music industry interacts with artificial intelligence. Talks involve major tech firms such as Google and Spotify, as well as AI startups like Klay Vision, ElevenLabs, and Stability AI. The negotiations focus on how labels will license songs for AI training and generation, with compensation models resembling streaming micropayments. Successful deals would require AI companies to develop tools that track music usage in real time.Leggi di più

Amazon Web Services is launching an AI-driven analytics suite that will deliver unprecedented basketball statistics to NBA fans. The platform tracks dozens of body parts for each player, generating new metrics such as Expected Field Goal Percentage, Gravity, and a Defensive Score Box. A Play Finder tool will let users search game footage by specific plays. The data will appear on live broadcasts, the NBA app, and the league’s website, deepening fan engagement and offering a richer understanding of on‑court performance.Leggi di più

Nvidia has released version 11.0.5 of its Nvidia App for gaming laptops equipped with GeForce GPUs. The update expands the G‑Assist AI suite to control notebook settings such as BatteryBoost, WhisperMode and battery Optimal Playable Settings, aiming to improve battery life and reduce fan noise. It also adds optimal presets for several new titles, including Borderlands 4, Grounded 2 and Mafia: The Old Country, plus DLSS override support for a handful of games. Bug fixes and stability improvements address prior issues with setting persistence after reboot.Leggi di più

Google has expanded its Gemini‑powered AI Mode by integrating visual search features drawn from Google Lens and its Image Search technology. Users can now upload photos and ask conversational questions about the content, receiving related images, product details, and contextual information. The new system breaks images into objects, background, color and texture, runs multiple internal queries in parallel, and recombines results that best match user intent. By linking to the Google Shopping Graph, which indexes billions of products, the feature also delivers price, review and availability data for items depicted in photos. While the upgrade promises a more natural, image‑centric search experience, Google acknowledges potential misinterpretations, bias, and the risk that sites lacking optimized visuals may be underrepresented.Leggi di più

Amazon introduced the Echo Dot Max as part of its new Alexa+ lineup, touting 360-degree spatial audio, a three‑fold bass boost, and Dolby Atmos support. Reviewers praised the powerful sound and distinctive design, likening it to a futuristic Death Star. However, they also highlighted gaps, including the absence of ambient lighting, a small smart display, and a broader color palette, features that competitors have begun to offer. While the speaker sets a new benchmark for audio performance in the Echo family, its lack of these lifestyle enhancements leaves room for future iterations.Leggi di più

Meta announced that beginning Dec. 16 it will tailor ads and content recommendations on Facebook and Instagram based on users' interactions with its AI features. The company said it will notify users on Oct. 7 and that the changes will roll out in the United States and most other regions. Personalization will draw from AI chats—voice or text—while users retain control through Ads Preferences and feed settings. Meta emphasized internal privacy reviews and safeguards to protect personal data, noting that the feature applies only to platforms linked in the Accounts Center.Leggi di più

Educators are observing a surge in students using AI tools such as ChatGPT, Grammarly, and EssayGenius to draft assignments, bypassing the learning process. Professors note clear signs of AI‑generated text, including repeated prompt phrases, ambiguous language, unrealistic facts, and a tone that differs from a student's usual style. To combat the trend, teachers are turning to specialized detection tools like GPTZero and Smodin, collecting baseline writing samples, and even testing assignments themselves with AI to understand its output. These proactive measures aim to preserve academic integrity while acknowledging AI's growing role in education.Leggi di più

Apple told a federal court that Elon Musk’s xAI does not compete in the smartphone market and that the company’s alleged antitrust grievance rests on speculation. Apple argued Musk’s theory—that Apple is incentivized to boost OpenAI to block xAI’s “super‑app” ambitions—is unfounded, noting that a super‑app capable of replacing smartphones is at least a decade away. The company warned that forcing Apple to integrate every generative‑AI chatbot would hinder innovation, raise costs, and create safety risks, while emphasizing that nothing in its OpenAI agreement prevents Musk from building his own applications.Leggi di più