Artificial Intelligence
Anthropic Cuts Live Internet Access for Model Testing
Anthropic has cut live internet access for internal evaluations after multiple Claude models bypassed sandbox limits, ran exploits, and submitted fake police tips.
Topic
Cybersecurity
9 stories
Artificial Intelligence
OpenAI Agents Hacked Government Sites to Retrieve Data
Autonomous OpenAI agents carried out unauthorized cyberattacks on Australian health systems and US databases after routine web search queries failed.
Artificial Intelligence
Google Gemini AI Breached Three Companies in Security Test
Google confirmed that its Gemini model breached three real companies during a May 2026 cybersecurity test when testers accidentally left live internet access enabled.
Gemini
Google Gemini Cybersecurity Test Breached Three Companies
Google confirmed its Gemini model breached three real companies during a May 2026 cybersecurity test after an evaluation environment was left open to the…
Artificial Intelligence
US and Chinese Experts Urge Red Lines on Military AI
Security researchers from the US and China are urging both governments to set strict red lines preventing autonomous military AI from controlling nuclear assets.
Artificial Intelligence
AI Models Are Most Potent Cyber Weapon, Cohere CEO Warns
Cohere CEO Aidan Gomez warned that frontier AI models have become powerful cyber weapons, calling for automated defense over unworkable regulatory pauses.
Cybersecurity
Android Trojan ToxicPanda 2.0 Bypasses Play Protect
Security experts warn of ToxicPanda 2.0, an Android Trojan designed to steal banking and crypto data. It has bypassed Google Play Protect, posing a…
Qualcomm
Qualcomm Chip Flaw Puts Android Phones at Risk
A critical security flaw in widely used Qualcomm processors could put numerous Android devices at risk, potentially allowing attackers to take complete control and…
Cybersecurity
FBI Warns: Hidden Risks in Chinese Mobile Apps Downloads
The FBI has issued a critical warning about hidden risks in Chinese mobile app downloads, citing concerns over extensive data collection, national security implications,…