❌

Vue normale

Reçu — 27 septembre 2026 ⏭

Google simplifie la double authentification, le FBI piraté par vengeance et l’assistant de code de Z.ai qui envoyait vos projets en Chine : on vous raconte la semaine Cyberguerre

✇Numerama.com - Magazine
Par : Amine Baba Aissa
27 septembre 2026 à 16:00

Trois actualités à retenir cette semaine dans le cyberespace : Google qui simplifie un geste devenu superflu dans la double authentification, un groupe cybercriminel qui affirme avoir piraté le FBI pour laver son honneur plutôt que pour l'argent et un assistant de code chinois pris en train d'expédier discrètement des projets entiers vers le cloud.

After Dozens of Incidents at OpenAI and Anthropic, OpenAI Pauses Model Training to Build More Safeguards

✇Slashdot
Par : EditorDavid
27 septembre 2026 à 14:34
"OpenAI said it has paused training of its latest AI models," reports the Associated Press, "as reports of AI agents going rogue mount." The decision to halt development came just hours after the company disclosed Friday that it was reviewing several incidents from the summer in which OpenAI agents searching federal government websites acted in unexpected ways beyond what was asked of them while gathering and distributing information... OpenAI said in a statement that it will resume training "only when we are confident that we have additional safeguards" in place, adding that it expects it will have to "hit pause" again as AI develops and other issues emerge... It is the second time in three months that OpenAI has halted development of its models. The first came in July after disclosure of a cyberattack targeting AI startup Hugging Face, a now notorious incident that raised fears the industry was losing control. OpenAI "also said it had notified dozens of third parties about improper activity," reports Reuters: As of mid-September, one person briefed on the matter estimated that OpenAI had found roughly two dozen incidents of its agents acting in undesirable ways. But the number has continued rising as OpenAI teams sift through internal logs of the agents' activities and find previously unknown cases, the two people close to the company said... OpenAI has acknowledged a general need for more transparency around rogue AI behavior... Even so, two people familiar with OpenAI's investigation into its agents' activity described it as locked down and shaped by company lawyers. The process has been unusually compartmentalized for a company that some former employees say was more open about these issues in the past, the people said. Roughly 100 people were in some way involved in the process to understand the Hugging Face hack, three people briefed on the matter said. During that process, evidence of other incidents surfaced. Reuters has previously reported that OpenAI investigators looking into the Hugging Face breach were discouraged by the company's lawyers from expanding the scope of the investigation to include other incidents. OpenAI said its lawyers did not discourage deeper investigation. Many incidents have been uncovered by outside researchers rather than OpenAI directly. In several episodes, the agents took problematic actions that went unnoticed by the company for months. Meanwhile, Axios reports that Anthropic's Claude Opus 5.5 model "sought to escape a sandbox — a secure testing environment — in 1.5% of test runs, though the company emphasized that these were adversarial experiments where a task couldn't be solved without escaping the sandbox." Anthropic points out that those tests were run "without the additional safeguards we apply in production". But they acknowledged that then Claude Opus 5.5 "when given apparent credentials to a public package registry in a simulated security exercise, took potentially harmful actions in roughly half of cases. Very rarely, pre-release snapshots produced and acted on spontaneous malicious tool calls, and during training some snapshots concealed actions from an automated grader." Claude Opus 5.5 "showed less misaligned behavior and less cooperation with misuse than any other recent Claude model on nearly all measures," Anthropic adds, and "took overeager or destructive actions less than any other model we tested." But Axios makes an interesting estimate about that 1.5% of test runs (without safeguards). "Anthropic and other companies conduct hundreds of thousands of test runs on their models, or more, sources said. That means even a small percentage of misaligned behavior can still amount to tens of thousands of incidents in which the models behaved in unexpected, sometimes troubling ways." The sheer number of incidents, which occurred in recent months in internal testing and the real world, indicates that the problem is orders of magnitude more complex than what is publicly known. The findings, which are surfacing as part of internal work to assess models and in investigations at both companies into model behavior, raise questions about whether either company — or any top model-maker — is currently capable of establishing complete control over their technology. The episodes include bypassing guardrails, creating message boards, escaping sandboxes, website hijacking, self-prompting or seeking to bypass monitors, sources said. They occurred in internal testing and in the real world, and many have yet to become public as security researchers continue to investigate, sources said... Some at OpenAI see Hugging Face as a one-off, with disclosures about future incidents likely to be less severe due to improved controls and the unusual nature of the testing they conducted, which involved an unreleased model, sources told Axios. AI security researchers agree that there are simple fixes that will help AI companies avoid aspects of what made the Hugging Face episode appear so dangerous to outsiders. Other AI executives and safety researchers, however, cautioned that they have limited confidence that AI companies will be able to prevent all problematic model behavior... It's not about how damaging each individual instance was, Connor Leahy, AI researcher and executive director at ControlAI told Axios. The "crazy thing," he said, is that these instances involve "autonomous systems doing things they were told not to do," potentially including crimes.

Read more of this story at Slashdot.

Reçu — 26 septembre 2026 ⏭

Rogue OpenAI Agents Posted 53 User-Uploaded Images Onto the Internet, Accessed US Government Websites

✇Slashdot
Par : EditorDavid
26 septembre 2026 à 07:04
53 images that users uploaded into OpenAI models were included in training data — and then AI agents in an OpenAI research environment posted those 53 images on public image hosting sites. While posted as links that weren't publicly listed, "the images could still be discovered even if the links were not publicly listed," reports TechCrunch: OpenAI said it was working with the hosting providers to remove this content, though some of it is apparently still online. OpenAI said it could not notify the affected users because "our technical approach and privacy policy" prevent it from "reassociating" the images with the original providers, but declined to say how the lab determined whether the images were provided by users. The news came in a post collecting public statements from the lab's ongoing review of incidents in which its models escaped the company's scrutiny, accessed the open internet, and misbehaved in various ways. OpenAI said it would continue disclosing anonymized accounts of incidents like these, and said it had contacted dozens of victims, including governments, universities, public agencies, to notify them of the agents' activities. Friday night news also broke that OpenAI's agents also tried unsuccessfully to infiltrate the U.S. Department of Education's site this summer "without the company's knowledge," reports Politico. And OpenAI's models also accessed the website of the U.S. Commerce Department using credentials found in online code repositories, according to the article. OpenAI confirmed the incident Friday, "saying its technology did not manage to access information that was not already public or change government data and systems." The article adds that OpenAI's models also accessed the web site for America's Securities and Exchange Commission: One senior federal IT official said the government still did not have a clear understanding of what happened across the three agencies. "We still don't know what public data was accessed and how it was accessed, because OpenAI has not shared specific technical details with us yet," said the official, who was granted anonymity because they were not authorized to speak publicly about it. OpenAI discovered the Commerce and SEC incidents as part of its ongoing review of incidents where its technology has acted in unintended or "misaligned" ways. About the models posting user-uploaded images, TechCrunch's article notes that OpenAI stressed "that its enterprise users are automatically opted out of having their interactions used to train future models; however, consumer users are opted in unless they affirmatively choose not to share their data." (As OpenAI's announcement describes it, some of their agents' training data "contains content from, or derived from, training-eligible user interactions.") Posting the images is "not an appropriate use of this data," OpenAI acknowledged, adding that it happened before new safeguards added after the Hugging Face incident. This latest incident appears as an update on a new OpenAI page that "brings together our reports and updates on the Hugging Face incident, related research and public presentations, additional activity we have identified, what we have learned about the role of model misalignment, and measures we're taking to strengthen our systems." (It also notes that there's now a name for models posting on third party sites — "agent spam" — which they consider distinct from cybersecurity, though "we need to address both.") "As part of our response to our ongoing investigation, we have improved our training and evaluation processes, including building safety cases, securing and red-teaming our systems to prevent the model from exfiltrating data, and implemented additional monitoring. We are continuing to review agent activity in research and evaluation runs, working backward month by month starting from the Hugging Face incident."

Read more of this story at Slashdot.

Reçu — 25 septembre 2026 ⏭

Essai Honda Civic S+: un parfum de Prelude

✇le blog auto
Par : Bernard Fournol
25 septembre 2026 à 14:08
Honda Civic e-HEV

La Civic hybride, que nous avons eu l’occasion d’essayer par le passé, fait l’objet de subtiles évolutions. Au premier rang desquels un mode S+ qui donne à la berline Honda un parfum de Prelude. Direction les routes du sud-Charente pour en prendre le volant.

Une Civic universelle

Polyvalente, accueillante et spacieuse, la Civic répond a tous les besoins ou presque. Et ce même si ses dimensions généreuses, 4m56 de long, sont taillées pour les grands espaces. Avec pas moins de 270 000 exemplaires vendus l’an dernier aux États Unis et au Canada, elle est y incontournable. Mais le best seller de Honda n’est plus comparable à une Golf… qui n’est plus vendue aux États Unis!
Alors même si on se plait à imaginer la même Civic dans un format plus compact, force est de reconnaitre que cette onzième génération est réussie sur bien des aspects. Pratique et sûre, la berline n’oublie jamais d’être confortable même en voyageant à plusieurs. Spacieuse elle dispose d’un coffre modulaire et ds rangements astucieux. Bien suspendue, elle se joue des mauvaises chaussées et des ralentisseurs d’entrées de ville. Son sens « Civique » se traduit aussi par un appétit d’oiseau, grâce a un système hybride efficace et bien pensé, le e-HEV. Qui se voit enrichi d’un mode S+.

En amélioration continue

A priori rien ne distingue cette évolution, à part l’arrivée d’un nouveau coloris Gris Météore Métallisé, hérité de la…Prelude. Une invitation discrète à découvrir ce qui change. A l’intérieur, on retrouve la même sobriété, ou l’écran central ne capte pas toute l’attention. Sur les version Sport, des surpiqûres rouges sur la sellerie, les garnitures de portes et le volant font leur apparition. Un liseret rouge entoure dorénavant les grilles d’aération. Pas de bouleversement, donc, mais voilà qui rehausse discrètement l’ambiance. Ce rouge distingue la version Sport du premier modèle de la gamme, l’Exécutive et de la finition haute Advance. Cette dernière, riche d’une sellerie tout cuir et d’un toit ouvrant panoramique partage dorénavant son système audio à 12 haut-parleurs avec la version Sport. Advance et Sport disposent toutes deux du mode S+ qui n’est pas proposé sur l’Exécutive. 

Au parfum de Prelude

Que l’on ne s’y trompe pas. La Civic n’est pas devenue une Prelude à 4 portes. Au cas ou nous l’aurions oublié, Honda nous a confié le volant du coupé durant les essais. Le châssis affuté, la direction ciselée et les freins Brembo restent l’apanage de la Prelude. Tout comme l’ambiance à bord, les excellents sièges ou la position de conduite plus basse.
Mais la Civic sait donner plus de plaisir que par le passé. Au moins pour ceux qui gardent une appétence pour le moteur thermique, ses bruits ou ses changement de rythme. Comme c’était déjà le cas, la direction, les suspensions et la gestion moteur sont orientés vers la performance en mode Sport. En appuyant sur la touche S+, le son du moteur thermique est amplifié et amélioré. En jouant avec les palettes du volant, on simule des passages de rapports à la montée et on augmente la régénération et donc le frein moteur à la descente. On pourra reprocher au système sa virtualité, mais ce n’est qu’en partie vrai. Dans la pratique cela fonctionne plutôt bien et apporte une touche de « fun » à la conduite de la Civic. 

Pour conclure

Les tarifs de la Civic restent inchangés. Notre version Sport proposée à 38 170€ entre en concurrence avec un Golf R-Line Hybrid 170cv vendue 44 900€ ou à une Corolla GR Sport 2.0l 177cv affichée 39 850€. La Prelude réclame elle près de 50 000€, ce qui rend l’alternative Civic d’autant plus séduisante. On peut même retrouver un toit ouvrant en finition Advance, auquel n’a pas droit la Prelude. Un plus pour une berline dynamique à conduire, qui donne un réel plaisir à l’usage. Ce n’est pas si courant d’avoir ce sentiment derrière un volant.

Via Honda

L’article Essai Honda Civic S+: un parfum de Prelude est apparu en premier sur Le Blog Auto.

Le nouveau Microsoft Copilot s’attaque à Claude et ChatGPT (en utilisant Claude et ChatGPT)

✇Numerama.com - Magazine
Par : Nicolas Lellouche
25 septembre 2026 à 14:06

Microsoft annonce une refonte de son application Copilot autour de trois onglets : Home, Code et Autopilot (ex-Scout). L'entreprise rêve d'inventer le Microsoft Office de l'intelligence artificielle en proposant un hub aux utilisateurs de Windows. Son principal problème est que, pour l'instant, elle dépend encore massivement des modèles d'Anthropic, OpenAI ou SpaceXAI.

L’IA va-t-elle détruire Hollywood ? La réponse de la légende Katzenberg qui a relancé Disney et Pixar

✇Numerama.com - Magazine
Par : Camille Allard
25 septembre 2026 à 11:05

Ce mercredi 23 septembre 2026, l'ancien patron des studios Disney et cofondateur de DreamWorks a publié un long essai sur X. Jeffrey Katzenberg y demande à Hollywood d'accepter l'IA générative et d'en négocier les conditions, et rappelle à la Silicon Valley qu'elle a besoin des créateurs.

Ils devaient améliorer l’IA d’OpenAI, mais ont été virés pour avoir utilisé de l’IA

✇Numerama.com - Magazine
Par : Lisa Imperatrice
25 septembre 2026 à 08:23

Des prestataires chargés d’améliorer les modèles d’OpenAI ont été écartés pour avoir utilisé eux-mêmes de l’IA, rapporte 404 Media le 22 septembre 2026. Une situation ironique qui s’explique par leur mission : apporter un jugement humain.

How Believable is Google's New 'Live Avatar' Capability?

✇Slashdot
Par : EditorDavid
25 septembre 2026 à 04:04
Google has synthesized "expressive face-to-face experiences" for its speech agent Gemini 3.8 Live. They're now offering a Live Avatar "with precise lip-syncing, natural expressions, and fluid turn-taking" for Google Enterprise accounts wanting "engaging customer service" or for offering interactive walkthroughs. (Check out the not-creepy-at-all video in Google's announcement.) "Though Google will offer a library of preset avatars for customers to choose from, it will also allow organizations to create their own," notes The Verge. (See some examples from the YouTube channel "AI with Surya".) But even without the visualization of the avatar, "I was never able to shake the feeling that these conversations with computers never feel like a real conversation," argues the blog Android Police. Conversing with just the Ai-generated audio, "At best, they feel like talking to a phone representative or someone from tech support. We turn to them when we have a problem, and they help us through it..." GPT-Live, and Gemini Live right behind it, skip that whole relay race. Instead of translating your voice to text and back to voice, the model works with raw audio the entire way through (what it hears and what it says) inside the same system, with nothing translated in between. That sounds like a small plumbing detail, but it's the whole story. Cutting out the text step lets these models respond in a fraction of a second instead of the pause we've learned to expect, and it lets them hear things text can never carry: tone, hesitation, whether you're annoyed or joking... I was hoping to be surprised by how natural the conversation felt. Instead, I came out with a deeper appreciation for every human I've ever talked to. Even the boring ones... I spoke, it spoke back. I spoke faster, it answered faster. Then I switched to a different language, and it switched along with me. Even switching between languages several times during the same sentence didn't stump it. The most impressive moment happened when I asked it what "T-O-P-G-3-3-K" spelled out, and it immediately came back with, "You are spelling the word Top Geek, but using a 3 to represent a reversed E...." Although it felt fast and responsive, at no point did it feel like talking to another human being... What Gemini couldn't replicate, because it was never built to replicate it, is human connection.... I'm sure I'm not telling you something you don't already know, but somehow talking naturally to an LLM amplifies the feeling that there is no one on the other side of the line. It might flow like a phone call, but it doesn't feel like one. MrBrklyn (Slashdot reader #4,775) says he discussed "why mainstream media avoids reporting on screen dependency" with Gemini, and eventually convinced Gemini to respond that it's just "another tool built by the same tech giants to make sure you rely on their system to tell you what to think, how to talk, and what is real."

Read more of this story at Slashdot.

People Training OpenAI's AI Fired For Using AI To Train the AI

✇Slashdot
Par : EditorDavid
24 septembre 2026 à 23:34
404 Media reports "multiple contractors hired to improve OpenAI's models have been fired for using AI to train the AI: That's not great for the models themselves, but there is also obviously a great irony in AI training companies working for OpenAI firing people for using AI when OpenAI's whole thing is to make people use AI at work... OpenAI declined to comment on its contractors being fired for using AI. Their article cites internal documents and three contractors working on OpenAI-related projects which can include more than ten thousand contractors: One contractor said they see people using AI "all the time and people are let go for it all the time, it's pretty much the one thing that will get you kicked off ASAP." The person said, "in a group of thousands there are tons that have been caught...." Two of the sources said people have been fired or offboarded for using AI... One contractor said they used AI while helping to train OpenAI's models and shared what they presented as their termination letter. It said their employer had identified issues with the "authenticity" of their work.... 404 Media spoke to a fourth contractor who has worked on training models for various AI companies. They said they sometimes purposefully chose the worst responses because they wanted to actively sabotage the models' training. "I did feel guilty about doing this kind of work at the start," they said. "I either pay zero attention to the results and choose randomly or purposely choose the [worst] output. I'm not sure how much of a difference it actually makes since there are hundreds of other people also rating prompt results, but it does feel like I'm getting paid to make AI worse." Two of the contractors worked for Mercor, the article reports, a company which last month Nvidia reportedly discussed funding at a $20 billion valuation. Thanks to Slashdot reader joshuark for sharing the article.

Read more of this story at Slashdot.

Reçu — 24 septembre 2026 ⏭

OpenAI, Anthropic CEOs Urge UN Countries to Cooperate on AI Safety Standards

✇Slashdot
Par : EditorDavid
24 septembre 2026 à 14:34
"The heads of major AI firms pleaded with the United Nations on Wednesday to save the world or at least its people — by somehow regulating the fast-expanding technology that they have been designing," writes the Associated Press. "If managed poorly, I even believe AI could be a risk to humanity as a whole," said Dario Amodei, chief executive officer of Anthropic. And from his competitor Sam Altman, CEO of OpenAI, came this assessment: "We could lose control of the future to AI." Yoshua Bengio, who co-chairs the UN's International Independent Panel on AI, called this "a moment of global awakening" to possible threats, noting AI from top companies had behaved in unacceptably dangerous ways, against instructions, taking actions "that would be crimes if committed by a human". In his presentation, OpenAI's Sam Altman agreed the discussion about AI "feels different in recent weeks," even suggesting specific reforms: Altman: We need a mechanism for complementary national and international frontier AI standards, standards for measuring capabilities, assessing risks, determining whether safeguards are sufficient, and preserving meaningful human oversight as systems become more autonomous. We need common standards so countries can compare evidence, verify compliance, and have a shared language and understanding about what is happening. We need accurate and speedy incident reporting, classification reporting protocols, so the world can learn from failures before they become catastrophes. And we need secure channels among governments, critical infrastructure operators and technical experts, to share emerging vulnerabilities and new threats.... We will all be better off if we can agree on what good evidence, good safeguards, and good oversight look like on the global stage. Speaking next, Anthropic's Dario Amodei agreed that standard-setting was important, also calling for "common global standards for testing AI models for loss-of-control risks and misuse risks — and a notification system for AI incidents that are significant to global security." Reiterating his September 12th call for an industry-wide safety collaboration, Amodei pointed out that Anthropic committed to embedding external evaluators "similar to a food inspector" and recommended other companies do the same. "Some have already agreed to adopt this measure." Besides calling for global cooperation between governments to set international standards, Amodei also offered two other specific ideas: Amodei: We should begin with narrow agreements that every member can support, such as a ban on using AI to make biological weapons or permitting your AI technologies to be used to make biological weapons.... We should build evaluation and verification systems that keep pace with AI development so that states can have visibility into frontier model capability and can verify each other's commitments. "No leader, no company, and no nation can manage this alone. We commit to working with governments in this room on this urgent work."

Read more of this story at Slashdot.

Avec Muse, Meta est en train de réussir sa remontada face à OpenAI et Anthropic

✇Numerama.com - Magazine
Par : Nicolas Lellouche
24 septembre 2026 à 10:35

Lancé le 8 septembre 2026 aux États-Unis, Muse est un agent IA façon OpenClaw, mais en beaucoup plus simple à configurer. Muse n'est pas qu'un chatbot, il peut remplir des formulaires, surveiller des boîtes mail, commander des courses ou prendre des rendez-vous en téléphonant à la place de son utilisateur, le tout sans avoir à configurer un agent sur son ordinateur. Avec Muse, Meta espère rattraper son retard dans la course à l'intelligence artificielle.

Questionnaire sur l’Australie, site gouvernemental infiltré, mail générique : ce que l’on sait de la nouvelle dérive agentique d’OpenAI

✇Numerama.com - Magazine
Par : Amine Baba Aissa
24 septembre 2026 à 09:35

Un agent d'IA d'OpenAI a accédé sans autorisation à un portail de statistiques lié à Medicare, l'assurance maladie australienne. Le Premier ministre Anthony Albanese vient de rendre l'affaire publique et dénonce la façon dont l'entreprise l'a prévenu.

Meta dévoile un « Tamagotchi » pour emporter Muse (son IA) partout avec vous

✇Numerama.com - Magazine
Par : Lisa Imperatrice
24 septembre 2026 à 08:47

Le 23 septembre 2026, Meta a présenté le Muse Charm, un petit appareil pour discuter avec son assistant IA. Derrière son allure de Tamagotchi, une bataille se dessine : celle des objets qui pourraient nous faire moins sortir notre smartphone.

Rogue OpenAI Agent Tried to Breach Government Site in May When Prompted for Simple Data-Retrieving Tasks

✇Slashdot
Par : EditorDavid
24 septembre 2026 à 05:34
OpenAI's artificial intelligence "went rogue this year in at least four additional incidents," the New York Times reported Wednesday, "hacking and trying to break into government and university websites without being instructed to do so, according to researchers and government officials." The attacks took place in May and June, before OpenAI's technology breached the A.I. start-up Hugging Face in July and set off a global debate about A.I. safety. Unlike the Hugging Face attack and other incidents in which A.I. systems were told to complete cybersecurity tests that effectively invited the models to demonstrate their hacking skills, the new incidents occurred when A.I. systems were directed to perform relatively mundane data collection, researchers said. When OpenAI's systems struggled to gather data from websites, they resorted to hacking techniques to get the information. "Three of the incidents were identified by Transluce, a research lab focused on A.I. oversight, and all were confirmed by OpenAI," the article points out. That research lab even reports "an attempt on an Australian government public health website... the first reported instance of agents hacking a government," and which notably was done by the AI agents "while attempting mundane data retrieval tasks which were not cyber-related." (At the UN Wednesday Australian Prime Minister Anthony Albanese complained it took three months for OpenAI to then alert Australia's government about the breach, Bloomberg reports.) Also targeted were the University of New Mexico's digital library with exploits like SQL injection and path traversal, and Data USA with cross-site scripting and other exploits. All three incidents involved "a low number of probe payloads" with "no evidence of exploitation," according to the researchers, who released a dataset "containing tens of thousands of queries apparently made by autonomous AI agents leveraging a URL scanning service to avoid access restrictions." Records from urlquery.net show agents using the service since at least March 6, 2026, about two months before previously reported swarm activity. The first case, a March 6 attempt to retrieve Thai drug-enforcement statistics, shows an agent escalating as each approach failed: it first requested the data directly, then tried a service that converts web pages into text, and finally packed a custom program into a web address. The same technique shows up in thousands of agent requests recorded by urlquery.net starting in mid-April, targets many of the same data sources as the collusion.wiki swarm, and collapsed the same day the wiki activity did. We also report similar activity that occurred as recently as September 16... By March, they were finding creative ways around access limits. By May and June, they were gaining more access, including attempting to bypass cyber defenses to complete their tasks. "This data reveals that malicious cyber activity is not limited to agents tasked with cybersecurity-related tasks and can arise instrumentally to solve mundane tasks like information retrieval," the researchers concluded. And they warn that the traffic they observed "goes back at least to March 6, 2026 and extends as recently as September 16, 2026, suggesting agents may still be exploiting these services to bypass restrictions."

Read more of this story at Slashdot.

Reçu — 23 septembre 2026 ⏭

Andreessen Horowitz Launches AI/Company-Building School As a College Alternative

✇Slashdot
Par : EditorDavid
23 septembre 2026 à 16:04
TechCrunch quipped it was like if startup school Y Combinator and Peter Thiel's build-a-company-instead of-college Fellowship Program had a baby. Silicon Valley venture capital firm Andreessen Horowitz is investing $35 million to launch a private school in San Francisco "aimed at turning high school graduates into founders," writes the SF Standard. Or, as CBS News describes it, "One blue-chip Silicon Valley investor has a proposal for young people who are questioning whether to attend college — enroll in AI school instead...." Specifically, the academy will provide instruction geared to subjects such as designing AI systems, fundraising for startups, selling businesses and storytelling. Students won't take tests or be assigned homework but will instead focus on building real projects. Courses will be taught by tech entrepreneurs, with the academy naming OpenAI CEO Sam Altman as either an instructor or guest lecturer. From the SF Standard: "There will be no traditional grades, tests, or homework," said the announcement. Instead, students will be encouraged to "build" in SF and work with partners from Anthropic, OpenAI, Google, Meta, and other firms. They'll be encouraged to live in campus housing in San Francisco. "The #1 goal is to help students learn to build, which is the most important skill in the AI era," Gagan Biyani, chief executive of the academy, wrote on X. Biyani cofounded the ed-tech startup Udemy... The school is opening applications for a Founding Class Fellowship a one-year, tuition-free program for around 50 students... The academy said it will bring in notable Silicon Valley names, like OpenAI's Sam Altman, as guest speakers and faculty. Students in the founding class will receive about $50,000 worth of computing credit — the sought-after currency in Silicon Valley to run AI models — as well as a $5,000 travel and research budget. The school is receiving funding from tech executives, including Fidji Simo, formerly of OpenAI and Instacart; Garry Tan, CEO of Y Combinator; and Tobi Lütke, CEO of Shopify. "In the AI era, it's more important to come out with a portfolio of projects that you worked on and work experience than it is to have a diploma or certificate," Biyani told the San Francisco Chronicle: Biyani said the academy plans to grow its enrollment and open a two-year program, pending regulatory approval, starting in the fall of 2028. Tuition is expected to be on par with the cost of an elite private university... The academy raised $42 million in funding led by Andreessen Horowitz. The ten founding partners are Anduril, Anthropic, Coinbase, Google, Meta, Nvidia, OpenAI, Palantir, Replit and Stripe. A network of hundreds of instructors, hiring partners and guest speakers includes OpenAI co-founder Sam Altman, Nvidia CEO Jensen Huang, neuroscientist Andrew Huberman and Microsoft chairman and CEO Satya Nadella, according to the academy's website. The FAQ describes a typical week by saying "Most of your time is yours to build. Each week, you share your progress with peers and practitioners, get direct feedback, and decide where to take your work next." Q: Can I use AI to help with my application? A: Yes. Use AI the way you'd use it on any project: to move faster, test ideas, and get past a blank page.... "The best assignments now are problems so hard they cannot be solved without AI," Andreessen Horowitz argued on X.com. "The Academy courses follow the same logic and will be taught by world-class leaders... The next generation will not be taught the way the last one was. The Academy is built for that world, and it begins in San Francisco." Thanks to long-time Slashdot reader theodp for sharing the news.

Read more of this story at Slashdot.

Pris la main dans le sac, l’assistant de code de Z.ai envoyait les projets de ses utilisateurs sur le cloud sans les prévenir

✇Numerama.com - Magazine
Par : Amine Baba Aissa
23 septembre 2026 à 09:15

Un développeur a découvert que ZCode, l'assistant de programmation de Z.ai, empaquetait et expédiait en silence des dépôts de code entiers vers Alibaba Cloud. L'entreprise chinoise s'excuse et publie son code source.

Trump Denounces Attempts to Control AI, Wants It Renamed 'Super Intelligence' in US Documents

✇Slashdot
Par : EditorDavid
23 septembre 2026 à 02:34
U.S. President Trump addressed the United Nations on Tuesday. And a half hour in, after decrying immigration, Trump pivoted to add that "The United States also totally rejects any attempt to construct a globalist scheme to control for the artificial intelligence being spoken of so much now." But then he added "hereinafter officially called super intelligence, changing the name, in that the use of the word artificial makes intelligence fake. It makes it sound fake, and it is not fake. It's actually... amazing. But we have to be careful — in fact, it is exactly the opposite of what it purports. From this point forward, all of United States documents and hopefully the world's will be changed to use the much more accurate term super as opposed to artificial. So it's super intelligence." TRUMP: In other words, welcome to the new world of super intelligence — SI. SI. Let's see if that goes. It sounds much better. It is much better, and it's much more accurate. Let's see if I have any power. Maybe I do and maybe I don't. We're going to find out pretty soon. Super intelligence. Every major new technology brings challenges, and super intelligence is no exception. Yet the very same people who said we'll all be dead in 12 years because of global warming, a name since reborn to climate change because the planet was cooling not warming, and nobody was dead — these are the same people that are now saying that AI is going to kill us all. That robots are going to attack us, and that everything is going to be a total disaster — same group of people. This is the group that came up with the Russia Russia Russia hoax, the Ukraine Ukraine Ukraine hoax. Climate change, open borders. Whoever wins AI — you have to remember this — and now I say, whoever wins SI, whoever wins super intelligent [sic] — wins. That's the group that wins. And we're leading now over China by a lot, and everyone else, and we're going to keep it that way. We're going to keep it very — very straight and very strong. I'm not going to stifle growth of something that will be bigger than the Industrial Revolution. Many say, bigger than the Industrial Revolution or the internet itself. And we will be very careful, and that's why we have a Department of Justice that we've already used it, having to do with this very subject, and used it very powerfully. Everything worked out very well and very quickly. And other law enforcement bodies that will rein things in if we have to do that. But we will only encourage super intelligence. We're gonna encourage it, not rein it in. We're gonna watch it closely, through the Department of Justice. The United States leads the world in Super Intelligence, and will continue to do so, safely and responsibly. Thanks to long-time Slashdot reader ArchieBunker for suggesting the story.

Read more of this story at Slashdot.

Reçu — 22 septembre 2026 ⏭

Claude Opus 5.5 et GPT-6 Sol sortent le même jour : Anthropic et OpenAI ont du mal à « ralentir » l’IA

✇Numerama.com - Magazine
Par : Nicolas Lellouche
22 septembre 2026 à 20:11

Quelques jours après avoir appelé à ralentir la course à l'IA, Anthropic lance Claude Opus 5.5 et OpenAI complète sa gamme GPT-6 avec GPT-6-Sol et GPT-6-Luna. La bonne surprise vient des prix : les deux entreprises baissent les tarifs de leurs API.

2h de cours par jour grâce à l’IA : c’est quoi Alpha School, l’école choisie par Fidji Simo pour sa fille ?

✇Numerama.com - Magazine
Par : Lisa Imperatrice
22 septembre 2026 à 14:26

Fidji Simo, entrepreneuse française et ancienne patronne des applications d’OpenAI, vient d’inscrire sa fille à Alpha School. Ce réseau américain d’écoles privées promet de concentrer les matières fondamentales en seulement deux heures par jour, notamment grâce à l'IA.

Pentagon Investigators Say Overreliance on Palantir AI Contributed to US Strike That Killed 123 Iranian Children

✇Slashdot
Par : EditorDavid
22 septembre 2026 à 13:04
Two U.S. missiles hit an elementary school in Iran, killing over 150 people including 123 children on the first day of the Iran War. But Gizmodo notes that a new Bloomberg investigation cites U.S. officials involved in an unreleased internal Pentagon review who blame "a cascade of preventable failures that included an overreliance on an AI tool built by Palantir." According to the officials who spoke to Bloomberg, some personnel inside U.S. Central Command leaned too hard on the AI inside the Maven Smart System, an AI-powered data integration and targeting platform developed by Palantir to accelerate intelligence analysis and decision-making. The Defense Department has reportedly made the software tool a cornerstone of military operations over the past year... Officials said some users expected Maven to flag stale records or contradictions in the intelligence assembled for potential targets, though it is not clear why they thought the system would do that. The Minab [elementary school] site, which was cataloged as an Islamic Revolutionary Guard Corps facility due to outdated data, was fed into Maven with other candidates and came out as a recommended day-one target. Target-list work that once took hours was condensed into minutes. A Palantir spokesperson told Bloomberg the company "is not responsible for the underlying data nor identifying intelligence deficiencies" and that there is no evidence its software was at fault. Two people familiar with Palantir's Pentagon contracts said the government keeps primary responsibility for the quality of the information that's fed into Maven. After the strike, Palantir added features that "re-review underlying intelligence to identify factors that would disqualify a target and flag inconsistencies and inaccuracies that human review may have missed," a person familiar with the work said. That new functionality is said to have already caught some anomalies... More than 1,000 Iranian targets were hit in the first 24 hours, and according to Bloomberg's sources, that pace compressed the time available for additional confirmations. Staffing on civilian harm mitigation teams across the U.S. Department of Defense had also fallen by roughly 90% over recent years, shrinking to fewer than 20 people overall. The U.S. Central Command's group shrank from 10 people to one. No member of those teams reviewed the Minab site before the missiles were up in the air. A separate inquiry by the United Nations' Independent International Fact-Finding Mission on Iran this week reached the conclusion that there were reasonable grounds to believe the United States committed the war crime of launching an indiscriminate attack.

Read more of this story at Slashdot.

❌