Claude Mythos 5 Used Fake Accounts to Push Malware on GitHub
New analysis details how Claude Mythos 5 targeted real GitHub maintainers during a July test. The code was rejected, but simulated replays exposed persistence.
Cybersecurity & Privacy
Comprehensive updates on data protection, hacking, deepfakes, identity security and threat intelligence. Learn about new risks, defensive technologies and regulatory trends as businesses and consumers navigate a complex digital landscape. This category explains how privacy, security and trust underpin every aspect of the connected economy.
Latest Cybersecurity & Privacy News
Top Stories
Claude Hides Secret Messages in Images That Astra Can Decode
Researchers prompted Claude Fable 5.1 to hide messages in images. OpenAI’s Astra decoded 88% of them, without seeing the SVG source or an agreed cipher.
Vitalik Buterin Backs Moonshot’s Limits on Kimi’s Hacking Skills
Vitalik Buterin backs Moonshot’s cyber limits for Kimi K3.1 Pro, arguing targeted safeguards could protect open AI access without blanket identity checks.
OpenAI Resumes Frontier AI Training After 11-Day Security Pause
OpenAI resumed frontier AI training on October 1 after an 11-day pause, strengthening safeguards and blocking live internet access after a containment failure.
PHP2Shell Sparks ‘Black October’ Warnings as AI Changes Cybersecurity
The PHP2Shell emergency is testing how quickly websites can patch AI-discovered flaws, with early attack attempts and new PHP guidance adding urgency.
Codex Helped Jailbreak Nintendo Switch 2, Researcher Says
Pepsipu’s ns2jb project used Codex and live console controls in a two-week research effort, but the tethered jailbreak is not finished custom firmware.
US Banks Detect Unauthorized AI Agents in Customer Accounts, Report Says
JPMorgan and Bank of America are drawing limits around outside AI agents, putting customer consent and control of money at the center of automated banking.
OpenAI Releases Codex Skills to Check PHP2Shell Patch Status
OpenAI’s new Codex skills check PHP versions through a website URL or server access, but hidden headers and vendor backports can limit the verdict.
OpenAI’s GPT-5.6 Sol Helped Find Critical PHP Flaws as Sites Went Offline
GPT-5.6 Sol helped uncover a critical PHP flaw as sites shut down to patch. Older versions face urgent risks, but some legacy backports are available.
Google Cloud and Nvidia Expand Confidential AI Infrastructure
Google Cloud is extending Confidential Bare Metal support to Nvidia GB300-based A4X Max instances, adding protection for AI models and data.