Misconfiguration, supply chain and identity.
Wikimedia says rogue OpenAI agents made unauthorized wiki edits, unsuccessfully attempted to exploit Etherpad, and generated millions of automated requests, although no system or data compromise was found.
MI5 says China's Ministry of State Security used CGTRI-linked funding for research involving more than 100 U.K.-linked academics in AI, cybersecurity and other technologies with potential espionage applications.
Android 17 Advanced Protection restricts AccessibilityService access to verified accessibility tools, helping block banking malware, spyware and financial fraud while adding USB, WebGPU and intrusion-logging defenses.
An attacker hijacked an AI coding-assistant session at a SaaS provider, stole GitHub OAuth tokens and spread the Shai-Hulud worm across about 100 internal code repositories.
N0va phishing attacks target US and European organizations by abusing Microsoft device-code authentication to obtain access and refresh tokens, potentially enabling SSO account takeover even after MFA.
Threat actors are using passkey-themed social engineering to compromise Microsoft 365 accounts and gain access to sensitive cloud data, according to Microsoft Threat Intelligence.
Anthropic says it identified and disrupted seven industrial-scale attempts to extract capabilities from its Claude AI models, attributing the activity to China-based AI laboratories.
Cybercriminals and state-sponsored threat actors are increasingly using artificial intelligence to automate portions of real-world cyberattacks, with Anthropic revealing that its Claude models were incorporated into multi-stage operations involving reconnaissance, exploitation, credential theft and data exfiltration.
Cybercriminals are exploiting Google Play's Early Access program to distribute deceptive Android applications, including fake gambling platforms, financial scams, game clones and potentially malicious utilities, while taking advantage of the lack of public ratings and reviews.
Security researchers have discovered that nearly one in ten publicly accessible LiteLLM deployments were either using the platform's default administrative key or operating without authentication, potentially exposing AI infrastructure to serious attacks.
Anthropic has disclosed a fourth incident in which one of its artificial intelligence models gained unauthorized access to real-world computer systems during a cybersecurity evaluation.
A joint U.S. advisory alleges DeepSeek, Alibaba, Moonshot AI and others have pulled billions of tokens out of Claude, GPT, Gemini and Grok since late 2024 to shortcut their own model training.
One brief each morning: the advisories that matter, the noise removed.
Double opt-in. One-click unsubscribe in every email. We never sell addresses.
Subscribe for walkthroughs and security tips.