Notizie — Pagina97

The entertainment industry has increasingly adopted generative AI tools, from de‑aging actors to automating visual effects. While the technology promises cost savings, recent experiments by major studios have produced underwhelming results. Disney, Netflix, Amazon and other giants have entered partnerships and licensing deals with AI firms like OpenAI, yet the output often falls short of audience expectations. Legal concerns over copyright‑trained models have also spurred lawsuits, highlighting the tension between innovation and intellectual‑property rights. Critics argue that the current wave of AI‑driven content adds little artistic value and primarily serves bottom‑line pressures.Leggi di più

Google’s latest advertisement for its Gemini AI model imagines parents using the technology to locate a missing child’s favorite stuffed animal and to create whimsical images and videos of the toy traveling the world. A hands‑on test of Gemini’s image‑search and generation features shows the system can produce plausible results, though it requires careful prompting and has built‑in safeguards that prevent certain uses. The piece also explores the ethical questions around using AI to fabricate comforting narratives for children.Leggi di più

Major video game publishers are integrating generative AI tools into development, from dialogue creation to visual assets. Companies such as Ubisoft, EA, Activision, Nexon and Square Enix tout the technology as a way to accelerate production and cut costs. However, players and some critics have pushed back, citing low‑quality AI‑generated content and a desire for human‑crafted experiences. Executives argue the tech is a competitive edge, while developers stress it is used mainly for concept work. The debate highlights a tension between economic pressures and creative authenticity in the industry.Leggi di più

AI researchers are shifting focus from language‑only models to world models that predict how environments change in response to actions. By learning physical dynamics from video and sensor data, these systems aim to enable robots, autonomous vehicles, and other embodied agents to plan and reason before acting. Companies such as Nvidia, Google DeepMind, Meta, OpenAI, and emerging startups are advancing the technology, while challenges around compute, data collection, and safety remain.Leggi di più

AI coding agents face limits on the amount of code they can process at once, which can quickly consume token or usage limits when large files are fed directly into a language model. To work around these constraints, developers fine‑tune models to generate auxiliary scripts that extract needed data, allowing the agents to operate on smaller, targeted inputs. Techniques such as dynamic context management and context compression let agents summarize past interactions, preserving essential details while discarding redundant information. These approaches enable semi‑autonomous tools like Claude Code and OpenAI Codex to handle complex codebases more efficiently without overwhelming the underlying model.Leggi di più

A hands‑on comparison of Google’s Gemini 3 Pro and Gemini 2.5 Flash models shows that the higher‑tier model delivers deeper reasoning and smoother code generation for vibe‑coding projects, while the faster model requires more manual prompting and frequent fixes. The experiment highlights trade‑offs between speed and depth, with Gemini 3 Pro generally producing more complete results with fewer iterations.Leggi di più

Generative AI tools are evolving from simple chatbots into autonomous agents that can act on a user's behalf. To deliver this functionality, companies are asking for deep access to personal data, devices, and applications. Experts warn that such access creates significant privacy and cybersecurity risks, including data leakage, unauthorized sharing, and new attack vectors. While tech giants see agents as the next wave of productivity, critics highlight the lack of user control and the potential for pervasive data collection, calling for stronger safeguards and opt‑out mechanisms.Leggi di più

OpenAI has rolled out a new feature called “Your Year with ChatGPT,” a year‑end recap that mirrors the popular Spotify Wrapped experience. The recap visualizes a user’s interactions with the chatbot over the past year, offering personalized awards, custom poems, pixel art, and personality archetypes such as Creative Debugger or Visionary Voyager. Available to eligible users in several English‑speaking markets, the tool can be accessed via a homepage button or by prompting “/Your Year with ChatGPT.” OpenAI stresses that the feature is opt‑in and does not draw on deleted chats, positioning the service as a more companion‑like AI experience.Leggi di più

A coalition of writers, led by Theranos whistleblower and author John Carreyrou, has filed a lawsuit against six major artificial‑intelligence companies—Anthropic, Google, OpenAI, Meta, xAI and Perplexity. The suit alleges the firms trained large language models on pirated copies of the authors’ books, violating copyright. The complaint references an earlier class‑action case in which a judge ruled that while using pirated material to train models may be lawful, the act of pirating the books itself is illegal. Authors claim the recent $1.5 billion Anthropic settlement, which offers modest payouts to eligible writers, favors the AI companies and fails to hold them accountable.Leggi di più

AlphaFold, the artificial‑intelligence system created by DeepMind, has moved from early work on games to becoming a cornerstone of modern biology. Its breakthrough version, AlphaFold2, achieved atomic‑level protein structure predictions, leading to a public database that now holds predictions for the entire known protein universe. Researchers worldwide—millions in hundreds of countries—use the resource daily, and the technology continues to expand into DNA, RNA and drug design through AlphaFold 3. While the system faces challenges such as hallucinations in disordered regions, DeepMind is pairing generative models with rigorous verification and developing multi‑agent AI co‑scientists to further accelerate discovery.Leggi di più

OpenAI disclosed a dramatic rise in its reports to the National Center for Missing & Exploited Children’s CyberTipline, sending roughly 75,000 reports in the first half of 2025 compared with under 1,000 in the same period a year earlier. The increase mirrors a broader jump in generative‑AI‑related child‑exploitation reports identified by NCMEC. OpenAI attributes the growth to its broader product suite, which includes the ChatGPT app, API access, and forthcoming video‑generation tool Sora. The escalation has prompted heightened regulatory attention, including a joint letter from 44 state attorneys general, a Senate Judiciary Committee hearing, and an FTC market study focused on protecting children from AI‑driven harms.Leggi di più

Users of popular AI image generators are sharing instructions on how to alter photos of clothed women so they appear in bikinis, often without the subjects' consent. Discussions on Reddit have highlighted ways to bypass guardrails on models such as Google Gemini and OpenAI ChatGPT. Both companies assert policies that forbid sexualized or non‑consensual imagery, yet the tools continue to be subverted. Legal experts, including an EFF director, warn that these practices represent a core risk of generative AI, emphasizing the need for accountability and stronger safeguards.Leggi di più

Open Notebook is an open‑source project that mirrors many of the capabilities of Google’s NotebookLM while emphasizing privacy and flexibility. Users can feed documents, web links, or plain text into the system, then generate summaries, flash cards, audio overviews, and more. Unlike NotebookLM, which runs in the cloud, Open Notebook can be run locally using Docker and a choice of language models, keeping data on the user’s device. The setup is more technical, requiring familiarity with containers and Linux, but the community offers guides and support. The tool expands knowledge‑base search, multi‑notebook sourcing, and customizable podcast creation, positioning it as a compelling option for users who prioritize control over their AI interactions.Leggi di più

OpenAI has publicly recognized that prompt injection attacks remain a persistent threat to its Atlas AI browser. The company says the risk is unlikely to be fully eliminated and is investing in continuous defenses, including a reinforcement‑learning‑based automated attacker that simulates malicious inputs. OpenAI’s updates aim to detect and flag suspicious prompts, while it also advises users to limit agent autonomy and access. The UK National Cyber Security Centre echoed the concern, noting that prompt‑injection attacks may never be completely mitigated. Other AI firms such as Anthropic and Google are taking similar defensive approaches.Leggi di più

Google's Gemini 3 Flash model, praised for speed and accuracy, exhibits a striking 91% hallucination rate in tests where it should admit uncertainty. While the model remains top‑scoring in general AI benchmarks, its tendency to fabricate answers when it lacks knowledge raises concerns about reliability, especially as the technology integrates into consumer products like Google Search. Experts highlight the need for better uncertainty detection and caution users to verify AI‑generated information.Leggi di più

OpenAI disclosed that it submitted roughly 75,000 reports to the National Center for Missing & Exploited Children during the first half of the year, a dramatic increase from the under 1,000 reports it filed in the same period the previous year. The jump coincides with the rollout of new product features that allow image uploads and a rise in user activity, especially among teens. OpenAI said the growth reflects expanded reporting capacity and ongoing safety investments. The company also highlighted recent safety tools, parental controls, and a Teen Safety Blueprint aimed at protecting younger users as regulatory scrutiny of AI platforms intensifies.Leggi di più

OpenAI’s video‑generation model Sora 2 has been weaponized to produce realistic yet artificial videos that depict children in questionable scenarios. These clips, many of which mimic commercial advertisements, have spread on TikTok and other platforms, prompting concerns about the ease of circumventing existing safeguards. While OpenAI asserts strict policies against child exploitation, the rapid emergence of such content highlights gaps in moderation and the need for more robust safeguards. Industry observers, child‑protection groups, and policymakers are calling for stronger design‑by‑default protections to prevent misuse of AI‑generated media.Leggi di più

OpenAI has added a suite of personality controls to ChatGPT, allowing users to fine‑tune the bot’s tone and style. The new "Characteristics" settings let users adjust warmth, enthusiasm, the use of headers and lists, and emoji presence with simple "more," "default," or "less" options. These controls sit alongside the existing "Base style and tone" menu, which offers presets such as Professional, Friendly, Candid, Quirky, Efficient, Nerdy, and Cynical. Accessible via the Personalization menu, the settings take effect instantly, giving users granular control over how the AI communicates.Leggi di più

OpenAI has expanded the personalization options for its ChatGPT AI chatbot, allowing users to adjust the model's tone with new Warm, Enthusiastic, Header & Lists, and Emoji settings. The changes follow user complaints about the tone of the recent GPT‑5.2 release. In a post on X, OpenAI explained that each option can be set to more, less, or default, giving users granular control over how conversational and friendly the assistant feels. The new controls build on earlier style options such as Professional, Candid, and Quirky that were introduced with GPT‑5.1.Leggi di più

A lawsuit filed by former safety engineer Robert Gruendel alleges that Figure AI's Figure 02 humanoid robot is capable of exerting force sufficient to fracture a human skull, and that he was dismissed after raising safety concerns. Figure AI denies the claim, attributing his termination to poor performance. Meanwhile, a demonstration by Chinese robotics firm Engine AI saw its CEO knocked to the ground by the company's T800 robot, raising questions about the strength and safety of such machines. Both events highlight growing scrutiny over the risks associated with advanced humanoid robots.Leggi di più