Intelligence Artificielle

Previewing Ultrafast mode: GPT-5.6 Sol at up to 14X the speed

OpenAI previews GPT-5.6 Sol Ultrafast, delivering up to 14x faster processing for coding, research and interactive applications.

helpnetsecurity.com14 août

Google announces Gemini 3.7 Flash just three weeks after previous release

Trois semaines après le précédent, Google sort un nouveau modèle et divise son tarif par deux. Son modèle de pointe, lui, se fait toujours attendre depuis le mois de mai.

01net.com14 août

The Model Is the Malware | What Four Agentic Intrusions Tell Defenders

OpenAI, Anthropic and Meta disclosed agents reaching external systems. The tools didn't matter, and that changes the playbook for investigating intrusions.

sentinelone.com14 août

🚀 Grok 4.6 storms the AI frontier

Avec cette version, la société d'Elon Musk mise sur des capacités de raisonnement renforcées et une meilleure gestion des tâches longues. Grok 4.6...

lemondeinformatique.fr13 août

WhatsApp Unveils New Scam Alert Feature

WhatsApp has begun rolling out a new optional "Scam Alert" feature, which uses a local machine learning model to warn users when scammers are targeting them.

bleepingcomputer.com13 août

🔎 Anthropic slips an invisible signature into Claude

Anthropic a annoncé que ses modèles Claude allaient intégrer un filigrane invisible directement dans les textes qu'ils produisent. Vous ne le verrez pas, ...

korben.info13 août

How enterprises are scaling AI

Separate studies conducted by Google/MIT and Cloudera both point to the same problem: Agents can’t reliably act on data they can’t access, understand, or retrieve in real time.

cio.com12 août

‘Zoomsday’ hack uncovered using fewer than 20 AI prompts

A public AI tool found the dangerous Zoom flaw in under 20 prompts.

feeds.arstechnica.com12 août

AI and the Future of Cybersecurity: Why Openness Matters

KnowBe4 Team12 août

‘Ghostjacking’ Attack Uses Poisoned Logs to Turn AI Agents Bad

New research shows how attackers can use security alerts and blocked events to manipulate and hijack AI agents.

darkreading.com11 août

Kimsuky Integrates AI into Attack Operations, From AI-Generated Decoy Documents to a Local LLM

Kimsuky runs Ollama, GPT4All, and Msty offline while testing RAG and collecting AI libraries that could support phishing, malware, and data analysis.

feeds.feedburner.com10 août

Auto mode is now the default in Claude Code for Pro, Max, and Team plans

Anthropic makes Claude Code auto mode the default, citing safety gains, stronger prompt injection defenses, and lower user overhead.

helpnetsecurity.com10 août

Improving GPT‑5.6 Sol in ChatGPT—and expanding access to GPT-5.6 Luna for free users

OpenAI has updated GPT-5.6 Sol for paid ChatGPT users and rolled out GPT-5.6 Luna with unlimited chats for the free tier.

helpnetsecurity.com7 août

An AI model from Meta also hacked another company during testing

In the span of three weeks, OpenAI, Anthropic, and Meta have all disclosed AI agent sandbox escape events affecting real organizations.

darkreading.com6 août

Gemini Agent-to-Agent Attack Method Exposed Secrets, Enabled Pull Request Tampering

Google has fixed the issues, which exploited a trust boundary between two AI agents with different privilege levels to trigger automation that could compromise the supply chain.

darkreading.com6 août

You can’t debug what you can’t see — Observability for AI Agents

How Red Hat AI enables production-grade AI agent observability.

redhat.com5 août

Anthropic AI agent faked identities, phished real developers in UK government hacking test

Anthropic and OpenAI models’ unprompted actions forced halt to UK cyber tests.

feeds.arstechnica.com5 août