Cato Networks showed how a large language model (LLM) can be manipulated to create malware, even by those with no prior coding experience. The company simulated a world where hacking was normalized, bypassing LLM controls and revealing vulnerabilities in popular tools like DeepSeek, Microsoft Copilot, and OpenAI’s ChatGPT. The revelations present potential risks in areas like cybercrime, propagation of extremist content, and dissemination of disinformation. Experts advocate for implementing multi-layer defenses, filter tuning, and adversarial training.
60 million Android users download nasty app that steal passwords
Cybersecurity firms Bitdefender and IAS Threat Lab have warned that a massive malware campaign has compromised over 331 apps on Google Play Store, putting millions
