Adobe has added a suite of AI-powered features to Acrobat that enable users to edit documents through natural language prompts, generate podcast-style audio summaries of their files, and create presentations by pulling content from multiple documents stored in a single workspace. The prompt-based editing supports 12 distinct actions: removing pages, text, comments, and images; finding and replacing words and phrases; and adding e-signatures and passwords. The presentation feature builds on Adobe Spaces, a collaborative file and notes collection the company launched last year. Users can point Acrobat's AI assistant at files in a Space and have it generate an editable pitch deck, then style it using Adobe Express themes and stock imagery. Shared files in Spaces now include AI-generated summaries that cite specific locations in the source document. Users can also choose from preset AI assistant personas -- "analyst," "entertainer," or "instructor" -- or create custom assistants using their own prompts.

Read more of this story at Slashdot.

Vous vous souvenez de la Toyota Aygo, qui partageait sa plateforme et ses moteurs avec les Peugeot 107/108 et Citroën C1 ? Aujourd’hui, il ne [...]

L’article Essai Toyota Aygo de 116 ch est apparu en premier sur Le Blog Auto.

San Diego Comic-Con changed an AI art friendly policy following an artist-led backlash last week. From a report: It was a small victory for working artists in an industry where jobs are slipping away as movie and video game studios adopt generative AI tools to save time and money. Every year, tens of thousands of people descend on San Diego for Comic-Con, the world's premier comic book convention that over the years has also become a major pan-media event where every major media company announces new movies, TV shows, and video games. For the past few years, Comic-Con has allowed some forms of AI-generated art at this art show at the convention. According to archived rules for the show, artists could display AI-generated material so long as it wasn't for sale, was marked as AI-produced, and credited the original artist whose style was used. "Material produced by Artificial Intelligence (AI) may be placed in the show, but only as Not-for-Sale (NFS). It must be clearly marked as AI-produced, not simply listed as a print. If one of the parameters in its creation was something similar to 'Done in the style of,' that information must be added to the description. If there are questions, the Art Show Coordinator will be the sole judge of acceptability," Comic-Con's art show rules said until recently.

Read more of this story at Slashdot.

Companies are spending vast sums on AI expecting the technology to boost efficiency, but a new survey from AI consulting firm Section found that two-thirds of non-management workers among 5,000 white-collar respondents say they save less than two hours a week or no time at all, while more than 40% of executives report the technology saves them upward of eight hours weekly. Workers were far more likely to describe themselves as anxious or overwhelmed about AI than excited -- the opposite of C-suite respondents -- and 40% of all surveyed said they would be fine never using AI again. A separate Workday report of roughly 1,600 employees found that though 85% reported time savings of one to seven hours weekly, much of it was offset by correcting errors and reworking AI-generated content -- what the company called an "AI tax" on productivity. At the World Economic Forum in Davos this week, a PricewaterhouseCoopers survey of nearly 4,500 CEOs found more than half have seen no significant financial benefit from AI so far, and only 12% said the technology has delivered both cost and revenue gains.

Read more of this story at Slashdot.

La période d’essai gratuite d’un an à Le Chat Pro de Mistral AI, déployée en février 2025 arrive bientôt à son terme, alerte Busyspider le 21 janvier 2026. Passé cette échéance, l’abonnement bascule automatiquement à 17,99 € par mois.

Dans une démonstration publiée sur LinkedIn le 16 janvier 2026, des chercheurs de Tenable ont révélé une faille de sécurité dans le nouvel outil d’OpenAI, ChatGPT Translate, rappelant les premières générations de garde‑fous appliqués aux agents LLM. 

Claude, l'IA d'Anthropic, peut désormais se connecter à des données de santé grâce à quatre nouvelles intégrations : Apple Health (iOS), Health Connect (Android), HealthEx et Function Health. La fonctionnalité est pour l'instant proposée en bêta aux États-Unis pour les abonnés Pro et Max.