Ultimate PocketIA
IA
Anthropic interdit la « torture » de Claude
Conscience en option Les IA n’ont « ni droits, ni sentiments, ni conscience », a récemment rappelé Mustafa Suleyman, le patron de Microsoft AI. Mais cela n’autorise pas pour autant les utilisateurs à torturer Claude, réplique Anthropic. Dans les nouvelles conditions d’utilisation des services d’Anthropic qui seront effectives le 12 novembre, l’entreprise interdit à ses clients d’« adopter un comportement abusif ou cruel prolongé et inutile envers [ses] modèles ». Cette condition, inscrite dans…
Lire l’article complet
IA
« OpenAI n’est pas une entreprise normale » : le licenciement de trois chercheurs en sécurité tourne au bras de fer public
Recevez tous les soirs un résumé de l’actu importante avec Le Récap’ C’est une affaire dont OpenAI a du mal à s’extirper. Le 1er octobre, on apprenait que l’entreprise américaine se séparait de trois chercheurs travaillant sur la sécurité de ses modèles : Jasmine Wang, Tomek Korbak et Mikita Balesni. Le Wall Street Journal rapportait alors qu’ils étaient soupçonnés d’avoir partagé des informations confidentielles avec une organisation spécialisée dans la sûreté de l’IA, et…
Lire l’article complet
IA
The Download: AI’s refusal problem and weight-loss drug side effects
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. We’re putting too much faith in AI’s ability to say no Today’s AI models are trained to refuse a vast number of prompts. If you ask your chatbot how to poison a colleague or how to tie a noose, chances are it’ll turn you down. But refusal fails, sometimes horrifically.…
Lire l’article complet
IA
OpenAI doubles down on decision to fire three AI safety researchers
The AI lab insisted the decision was about a ‘breach of trust,’ not raising safety issues. The AI lab insisted the decision was about a ‘breach of trust,’ not raising safety issues. OpenAI is standing firm on its decision to fire three safety researchers after an investigation found they committed “a significant breach of trust.” In a post on X on Friday, the company said Jasmine Wang, Tomek Korbak and Mikita Balesni were dismissed…
Lire l’article complet
IA
Google dévoile AI Edge Foresight : l’outil de prise de notes automatique gratuit et privé
Google lance AI Edge Foresight. Un logiciel de prise de note par IA entièrement gratuit qui fonctionne sur votre machine, sans passer par le cloud. Vos données restent donc sur l’ordinateur. Des IA qui prennent des notes de réunions ou de visioconférences à notre place, il y a en pas mal. Notion AI, Scribe, Granola… Dans la grande majorité des cas, les conversations sont captées puis envoyées à des serveurs distants avant d’être retranscrites…
Lire l’article complet
IA
Actualité : IA : Ecosia met un vent au français Mistral au profit de modèles open source et chinois
© Casimiro PT/Shutterstock - Le partenariat entre l'Allemagne et la France via Ecosia et Mistral est déjà terminé. En mai 2026, l'allemand Ecosia remplaçait l'étasunien OpenAI par le français Mistral comme fournisseur de modèles d'IA pour son moteur de recherche orienté écologie. Le 6 octobre dernier, lors d'une interview accordée à Politico, Christian Kroll, fondateur et directeur général d’Ecosia, a annoncé que ce partenariat appartenait désormais au passé. Pour expliquer sa décision, l'homme est…
Lire l’article complet
IA
50 ou 70 milliards ? Les chiffres d’OpenAI font vaciller les valeurs de l’IA à Wall Street
OpenAI aurait annoncé à ses investisseurs un revenu annualisé proche de 50 milliards de dollars fin septembre, contre 70 milliards évoqués auparavant, écart attribué à des méthodes de calcul. L’information a fait reculer le Nasdaq et pesé sur Nvidia, Oracle et Intel. Les revenus d’OpenAI ne seraient pas aussi élevés que certains investisseurs l’imaginaient, et Wall Street l’a fait payer aux valeurs de l’intelligence artificielle jeudi. Selon CNBC, OpenAI a indiqué à ses investisseurs…
Lire l’article complet
IA
We’re putting too much faith in AI’s ability to say no
Ever since people first seriously contemplated giving machines an intelligence modeled on our own, there has never been any question that they would, like us, be able to say no. The sci-fi canon is full of stories of robotic disobedience. Most of these capers are, of course, cautionary. But recently, the idea that AI shouldn’t do everything you ask has become something like a commandment. In 2021, a team at Anthropic wrote that large…
Lire l’article complet
IA
Apple Signs Hiring and Licensing Deal With AI Podcast Startup Huxe
Apple has reached an agreement to recruit employees from Huxe and license the AI audio startup's technology, according to a new European Commission filing. The filing says Apple will have the right to make employment offers to certain Huxe employees and receive a non-exclusive license to the company's intellectual property, so this isn't an outright acquisition of the company. Huxe was founded by former Google NotebookLM team members Raiza Martin, Jason Spielman, and Stephen…
Lire l’article complet
IA
Il est désormais interdit de harceler Claude AI
Recevez tous les soirs un résumé de l’actu importante avec Le Récap’ Faudra-t-il dire « bonjour » et « merci » à Claude pour avoir une réponse ? Le 8 octobre 2026, Anthropic a publié une mise à jour de sa politique d’utilisation qui fait beaucoup réagir sur les réseaux. Le point qui fâche : Anthropic bannit les comportements « abusifs ou cruels, persistants et gratuits » envers ses modèles. Dès le 12 novembre,…
Lire l’article complet
IA
Fired OpenAI safety researchers dispute their dismissals in open letter
They said company communications around the firing may have made colleagues 'afraid to speak.' Three OpenAI safety researchers have posted an open letter to OpenAI disputing their dismissals. Jasmine Wang, Mikita Balesni and Tomek Korbak wrote that the "very public manner" of the company's communications around the firings could have a "chilling" effect on employees at a time when "the world's safety depends on them." Last week, OpenAI dismissed three employees who allegedly shared…
Lire l’article complet
IA
Microsoft 365 Family subscribers will finally be able to share AI benefits
But OneDrive storage is being cut to a shared 2TB pool across all family members. But OneDrive storage is being cut to a shared 2TB pool across all family members. Microsoft bundled its AI-powered Office features into Microsoft 365 Personal and Family subscriptions last year, but it only allowed the primary account holder to access the AI benefits. Now, Microsoft is about to let Microsoft 365 Family and Premium subscribers share Copilot and AI…
Lire l’article complet
IA
Dépassée par son succès, Anthropic suspend son offre Claude for Startups
PR stunt Anthropic promettait d’offrir jusqu’à 1 000 dollars de crédits API et un an d’abonnement gratuit aux startups qui s’inscrivaient sur sa plateforme, après simple validation d’un formulaire. Trois jours plus tard, l’entreprise fait marche arrière, se disant victime de son succès. Trop beau pour être vrai ? Le 6 octobre dernier, Anthropic a annoncé l’extension de son programme Claude Startups, avec une promesse particulièrement séduisante : les entrepreneurs éligibles se voyaient offrir…
Lire l’article complet
IA
Roundtables: A Conversation With the Creator of AI-Designed Viruses
Available only for MIT Technology Review subscribers. Can AI design new life forms? In 2025, Stanford University PhD student Samuel King came up with a preliminary answer when he used a generative AI model to propose genetic blueprints for microscopic viruses. It isn’t yet an example of AI-generated life, but that could be next. Join senior AI reporter James O'Donnell as he interviews King about his work, being named one of MIT Technology Review's…
Lire l’article complet
IA
Anthropic launches free AI security scans for open-source projects
The new OSS Scanner service offers vulnerability reports from Anthropic’s “strongest models,” including Mythos. The new OSS Scanner service offers vulnerability reports from Anthropic’s “strongest models,” including Mythos. Anthropic’s offering to help open-source projects track down security vulnerabilities with a new service called OSS Scanner. It says open-source projects that opt-in will get “thorough, periodic security scans by our strongest models at no cost.” That could mean open-source projects get alerted about possible security…
Lire l’article complet
IA
Anthropic Says Users Can't Be Needlessly Cruel to Claude
Anthropic today updated its usage policy (via The Verge) to bar users from subjecting Claude to "sustained and needless abusive or cruel behavior." Anthropic says the abuse rule is meant to apply only in extreme cases, where users "repeatedly act cruelly" toward the AI models with "no discernible purpose." The rule does not apply to ordinary user frustration, pushback, dark creative themes, or model testing and research. Claude models are already able to end…
Lire l’article complet
IA
Anthropic bans 'sustained and needless abusive or cruel behavior' toward its AI models
The company is also tightening its election policy ahead of the midterms. Anthropic's annual usage policy update includes one eyebrow-raising change: a ban on "sustained and needless" cruelty toward its AI models. This comes in the shadow of a viral "AI torture chamber" project. 2026 sure is shaping up to be a wild one. "We've added a prohibition on sustained and needless abusive or cruel behavior toward our models," Anthropic's update reads. It describes…
Lire l’article complet
IA
Child safety group calls ChatGPT for Teens an unacceptable risk
Common Sense Media found the chatbot failed in five key areas. When OpenAI released ChatGPT for Teens this past August, child safety experts Engadget spoke to at the time expressed skepticism that the new mode would adequately protect young people from harm. Now, one of those groups, Common Sense Media, has published a formal risk assessment of the software, and deemed it an "unacceptable risk" for all children under the age of 18. "Some…
Lire l’article complet
IA
Fired OpenAI safety researchers dispute misconduct claims, warn of chilling effect
Jasmine Wang, Tomek Korbak, and Mikita Balesni, the three safety researchers that OpenAI fired last week, have published an open letter denying the firm’s claims that they mishandled sensitive information outside of established company procedures and warned that their dismissal signals a chilling effect that will have ripple effects across the company’s culture. “We have become concerned that internal and external communications around our firing have made our former colleagues afraid to speak and…
Lire l’article complet
IA
Ecosia quitte Mistral pour des IA chinoises, Arthur Mensch répond : « Désolé si le nucléaire est un vrai point de blocage »
Christian Kroll, fondateur et patron d’Ecosia, claque la porte à Mistral. Dans un entretien publié par Politico le 6 octobre, il annonce que son moteur abandonne le laboratoire français au profit de modèles à poids ouverts, y compris chinois. Le même jour, Mistral présentait Large 4, le plus gros modèle de son histoire. Difficile de faire pire comme calendrier. Ecosia, c’est ce moteur berlinois qui revendique plus de 20 millions d’utilisateurs et reverse ses…
Lire l’article completAI-generated political robocalls could flood your phone if FCC green-lights petition
TL;DR The FCC currently has rules in place restricting both unsolicited political robocalls, and those created with AI. It’s not facing a petition that would grant a special exemption, targeting upcoming midterms. One Commissioner warns this could hit people with a “a tsunami of robocalls.” How much do you love the FCC under the current administration? When the agency’s not banning our favorite robot vacuums, or drones, or the routers we need to get…
Lire l’article complet
IA
Ben Affleck is an AI nerd, and the internet is impressed
They cast the wrong lead in “Good Will Hunting,” apparently, because it’s Ben Affleck who is the secret genius, as it turns out. The actor who earlier this year sold his AI filmmaking startup to Netflix for a reported $587 million has gone viral this week for demonstrating his understanding of AI technology, neural networks, machine learning, and the like, across multiple video clips of his recent interviews. The movie star delved into his…
Lire l’article complet
IA
Popular AI leaderboard Arena nearly doubles valuation to $3.1B valuation in 10 months
Arena, which originated in 2023 as a research project at UC Berkeley that crowdsourced rankings of AI models, has raised a $200 million Series B round at a $3.1 billion valuation, it said on Thursday. This comes after the company said it reached $100 million in annualized run-rate revenue in June. The round was led by Lightspeed Venture Partners and Khosla Ventures, with Salesforce Ventures, 01 Advisors, Dell Technologies Capital, Endeavor Catalyst, a16z, Felicis,…
Lire l’article complet
IA
OpenAI’s revenue is reportedly $20 billion less than previously projected
A little over a week ago, it was reported that OpenAI’s annualized revenue was approaching $70 billion, a figure that would have made it competitive with Anthropic’s reported run rate. Now, however, the AI lab is said to have told investors that the real revenue is some $20 billion lower than that. The Financial Times reports that the company has told investors that its annualized revenue is “approaching $50 billion.” The $70 billion figure…
Lire l’article complet