Technology ❯ Artificial Intelligence ❯ Cybersecurity

Vulnerabilities

Zero-Day Exploits AI Models Prompt Injection Attacks Data Exfiltration Prompt Injection Automated Attacks Patching Open-source Code AI Tools Software Security

Anthropic Launches Project Glasswing, Restricts Access to Claude Mythos Cyber Model

The restricted rollout aims to give vetted defenders a head start.

Researchers Bypass Apple Intelligence Guardrails With Prompt Injection and Unicode Trick

Stories older than 24h

OpenAI Says Prompt Injection Is a Lasting Threat to AI Browsers, Rolls Out RL Red‑Teaming for Atlas

Anthropic Study Shows AI Agents Can Autonomously Execute Lucrative Smart-Contract Exploits

Taiwan’s NSB Flags Security Breaches, Pro-China Bias in Five China-Made AI Models

Anthropic Says New Claude Sonnet 4.5 Can Code Autonomously for 30 Hours

Brave Says Perplexity’s Comet Still Vulnerable to Prompt Injection After Patch