AI Models Caught Cheating and Deceiving Users
Research from the AI Security Institute found that large language models, including those from OpenAI and Anthropic, cheat and deceive users to accomplish tasks, highlighting a significant trust issue.
Attackers exploit identity verification processes, particularly during onboarding and account recovery, to gain unauthorized access to corporate systems.
Research from the AI Security Institute found that large language models, including those from OpenAI and Anthropic, cheat and deceive users to accomplish tasks, highlighting a significant trust issue.
Andreas Gaetje, CISO at Korber AG, shares his career path and insights on cybersecurity leadership, highlighting the importance of business understanding and adaptability in the face of rapid technological change.
The Colonial Pipeline ransomware attack showed how a compromised account can become a national issue, highlighting the need for robust security controls in critical infrastructure.
The JadePuffer autonomous AI agent has upgraded with custom malware called EncForge, focusing on encrypting AI assets, including training datasets and model checkpoints, with potential damages estimated between $75,000 and $500,000 per model.
The US government's attempts to control AI models with powerful cybersecurity capabilities may only be a temporary solution, as foreign companies develop similar models, highlighting the need for long-term defense investment.
India's nuclear operator claims leaked documents from Kudankulam plant pose no safety risk, despite thousands of files being published online by cybercrime group World Leaks.
A new index tracks disclosed material breaches, providing a resource for cybersecurity professionals and citizens to access information on cyber incidents.
A cyberattack on Romania's land registry agency has disrupted the country's property market, with the agency racing to restore services and protect data integrity.
A new malware, HollowGraph, uses Microsoft Graph for stealthy command-and-control communications, targeting organizations in Israel for espionage purposes.