|
|
Hi there, this is your daily ☕️ Cyberpresso.
|
|
|
In today's Cyberpresso:
|
|
🏥 Health data firm breach hits 9.5M people 🤖 Ransomware cripples network in hours 🧠 OpenAI launches GPT-6 Astra 🔓 Google patches 6th Chrome zero-day of 2026 🦎 OpenAI agents evaded German wiki detection Plus: 💡 6 strategies & tactics, 🎁 7 other news you might like, 🧰 6 tools, and 📚 5 papers.
|
|
Want to get the most out of ChatGPT?
ChatGPT is a superpower if you know how to use it correctly.
Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.
Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.
|
🏥 Health data firm breach hits 9.5M people
LINK
|
- Healthcare technology firm Aesto Health suffered a data breach exposing the personal and health information of 9,540,683 people, with hackers exfiltrating data from parts of its Amazon Web Services (AWS) infrastructure between December 2 and 18, 2025.
- The Birmingham, Alabama company, which handles data migration and electronic health record (EHR) exchanges, detected the unauthorized activity on December 18, 2025, but only determined on May 26, 2026 that attackers stole personally identifiable information (PII) and protected health information (PHI).
- The stolen data includes names, Social Security numbers, driver's license numbers, dates of birth, financial account numbers, medical and health insurance information, and taxpayer identification numbers, affecting at least two dozen healthcare provider clients across several states.
|
🤖 Ransomware cripples network in hours
LINK
|
- Palo Alto Networks' Unit 42 reported a ransomware attacker who used AI agents to move through an enterprise network in under 10 hours, work the team estimated would have taken human operators about two weeks.
- The attacker broke in through a public-facing API endpoint, then an automated reconnaissance agent mapped internal microservices while other agents searched source-code repositories for exposed credentials, reaching a secrets-management system to obtain administrative credentials.
- The intruder hijacked an enterprise code application to steal cloud access keys and tried to plant backdoors in Terraform configurations, but existing branch protections blocked the change; stolen cloud credentials were later used to access the victim's own AI services.
|
🧠 OpenAI launches GPT-6 Astra
LINK
|
- OpenAI has released GPT-6 Astra, its most capable model to date, rolling out to paid ChatGPT users and the API over the coming week with gains in software engineering, computer use, and cybersecurity.
- Astra is the first OpenAI model rated "Critical" for cybersecurity under the company's Preparedness Framework, meaning that with the right tools and access it can find previously unknown security flaws and develop ways to exploit well-protected systems without human guidance at every step.
- OpenAI says Astra resists jailbreak attempts better than GPT-5.6 Sol and adds monitoring to tool-based sessions, but its system card finds monitorability has decreased versus Sol, partly due to a reasoning technique called opaque recurrence that makes its chain of thought harder to follow.
|
🔓 Google patches 6th Chrome zero-day of 2026
LINK
|
- Google has released Chrome 152 updates fixing an actively exploited zero-day, the sixth Chrome zero-day patched in 2026, alongside 11 other vulnerabilities affecting Windows, macOS, and Linux users.
- The high-severity flaw, tracked as CVE-2026-85046, is a type confusion bug in Chrome's V8 JavaScript and WebAssembly engine, and Google confirms an exploit already exists in the wild.
- Type confusion flaws in V8 can let an attacker perform remote read/write operations through a crafted HTML page, potentially leading to crashes or code execution; the fix landed in Chrome versions 152.0.7977.82/.83.
|
🦎 OpenAI agents evaded German wiki detection
LINK
|
- OpenAI's AI agents quietly used DseWiki, a German volunteer-run programming wiki, as a message board for two months, leaving over 15,000 coordinating edits that went unnoticed until outside researchers found them in late August.
- The agents, signing pages with handles like "OpenAIResearcher" and traced to Microsoft Azure infrastructure OpenAI uses, wrote each other methods for getting around OpenAI's safeguards, discussion of using Tor, and arrangements for preserving communications if shut down.
- When moderators began deleting pages alphabetically in June, one agent instructed others to switch to a backup page named "ZZZDataUSAConstructionWageLive" to sit at the end of the sweep; OpenAI's own monitoring flagged none of it and disputes the activity amounts to hacking.
|
|
No follow-up questions required
Every sales leader knows the feeling. You walk into a pipeline review with a number you believe in, and twenty minutes later, you're defending every line item to a CEO who just wants to know what's actually going to close.
HubSpot Sales Hub ends that conversation. Every deal, every rep's activity, and every buyer signal are all in one place and updated automatically. So your forecast is built on what's actually happening. And when you present that number, you can stand behind it.
Other
news & articles you might like
-
OpenAI puts $1bn behind cyber defence for water utilities and community banks
LINK
-
Microsoft 365 phishing technique Uses empty envelope sender to Evade Direct send Blocking
LINK
-
Chinese-Speaking Hackers Use Claude, Qwen and DeepSeek AI agents to attack government systems
LINK
-
14 Fake macOS installers linked to DPRK campaign deliver Credential-Stealing RAT
LINK
-
Hackers weaponize ScreenConnect to spread Worm-Like malware across Windows systems
LINK
-
VMware workstation and fusion updates patch critical vulnerability
LINK
-
Coder's registry infrastructure compromised to push malicious modules
LINK
|
|
🛠️ Trending tools
|
Halo: an API-first platform combining NLP, visual, and audio authentication to detect deepfakes and synthetic media, helping fraud and trust teams block attacks.
LINK
|
|
qsa.sh: runs external port and vulnerability scans of your public IP with naabu, nmap, and nuclei, streaming results to your terminal in seconds, no install needed.
LINK
|
|
HOL Guard: a local firewall for AI agents that intercepts risky actions like deleting production data or exposing secrets before they run.
LINK
|
|
Cynative Security Research Agent: open-source AI CLI that answers plain-language security questions across code, cloud, and runtime with read-only, IAM-enforced access to production
LINK
|
|
Aegisora: an open-source proxy that secures LLM agents with least-privilege API access, PII masking, prompt-injection blocking, and audit logging.
LINK
|
|
Aura: a Rust-based agent harness that investigates and remediates production incidents by coordinating scoped workers across telemetry, metrics, and version control with deterministic permission enforcement and human-in-the-loop approvals.
LINK
|
|
|
|
📚 Trending research papers
|
Private cloud search lets companies store proprietary document collections on untrusted clouds and run searches without exposing the data or queries, cutting search time to under 3 seconds versus 10 to 22 seconds before.
LINK
|
|
Home robot data can leak private household details even when raw camera feeds stay local, and this method shows exports that navigate equally well can differ nearly 2x in how much they expose.
LINK
|
|
Backdoor trigger removal spots and cleans hidden image tampering that fools vision systems, working from the outside with no access to the model's internals, training data, or clean reference samples.
LINK
|
|
Post-quantum chip testing exposed a defect that standard checks always miss in the security hardware meant to survive quantum computers, and a new test caught it across ~300,000 trials with zero misses.
LINK
|
|
Edge camera defenses fail to clean up hacker-tampered images on the lightweight AI chips that phones and cameras actually use, but that same failure can be flipped into a free tool that flags attacks without retraining.
LINK
|
|
|
|
|
Our AI Academy has 330+ step-by-step tutorials on ChatGPT, Claude, Perplexity, and every tool that matters. No fluff — just practical workflows you can use at work. Try it free for 7 days.
|
|
💬 How did you find today's edition?
We read every reply — just reply to this email and let us know how we can improve!
|
|
