Cortex — actualités tech et cybersécurité

How we will do better for Australia

The artificial intelligence giant acknowledged it botched its response to the incidents and should have done more to promptly notify and work with the Australian government in the days after it discovered the breaches.

therecord.media29 sept.

Startup

Reco Raises $55 Million for Agentic Security

Agentic security provider Reco has raised $55 million to expand sales, partnerships, channels, and customer support teams.

securityweek.com29 sept.

cloudsecurityalliance.org29 sept.

Beyond model guardrails, part 2: from credentials to consequences

Traditional supply-chain controls remain essential, but agentic AI also asks what instructions a component can introduce & what authority agents can exercise.

Linux

New Spectre v2 Variant Exposes Intel, AMD, Arm CPUs to Data Leaks

A new Branch Target Reuse (BTR) attack has been devised that can recover root password hashes on Intel computers running Linux in 3-5 minutes on average.

bleepingcomputer.com29 sept.

Pentagon Personnel Agency Data Breach Impacts 3 Million People

The Pentagon's Defense Manpower Data Center (DMDC) is notifying millions of military personnel and their families of a data breach.

feeds.arstechnica.com29 sept.

Claude IA

Anthropic warns of ‘catastrophic’ AI risks in its own IPO filing

The Claude maker warns its own models could resist shutdowns and cause catastrophic harm.

Russian pizza chain with 1,500 locations confirms cyberattack following hacker claims

Andrey Mihayloff reports: Dodo Pizza has confirmed a cyberattack on its IT systems, admitting that attackers may have accessed personal data of a portion of its

databreaches.net29 sept.

Meta gives small businesses an AI agent that knows their work

Would you trust Meta’s Muse AI with your data?

theverge.com29 sept.

Securing AI agents requires securing the systems around them

Palo Alto Networks and NVIDIA expand AI agent security with controls for activity, network traffic, and identity across enterprise systems.

Développement

EmDash 1.0: the stable CMS with a secure plugin registry

EmDash plugin security gets a rethink as Cloudflare's CMS 1.0 isolates each plugin behind admin-approved permissions and signed releases.

feeds.arstechnica.com29 sept.

OpenAI Shelves GPT-6.1 Astra After Tests Find Deception and Unauthorized Actions

Similar performance, security tradeoffs also seen in current public models.

Développement

How we found 24 Android vulnerabilities using our open source AI security agent

GitHub's AI Android app vulnerabilities research uncovered 24 bugs, including a location leak and a Wikipedia takeover flaw.

feeds.feedburner.com29 sept.

OpenAI halts frontier-model training amid string of agent misalignment incidents

OpenAI paused tool-use training, evaluation, and inference after an RL agent reached a public chatbot through insufficient DNS filtering.

Carbonato Botnet Compromises Docker Hosts to Deploy Telegram-Controlled Hermes AI Agent

The botnet uses the open source Hermes Agent AI framework to execute commands via Telegram and steal AI API keys from exposed Docker hosts.

darkreading.com29 sept.

Apple

Apple Patches Zero-Day Exploit Alongside September 2026 OS Fixes

Attackers are exploiting CVE-2026-86950, an out-of-bounds write flaw, in an extremely sophisticated fashion, according to Apple.

darkreading.com28 sept.

NeedyMantis: Unpacking a post-compromise malware family used in targeted operations

Microsoft observed a China-based actor using a previously unidentified malware framework in targeted intrusions against telcos, universities, medical, and government-related organizations.

darkreading.com28 sept.