Hackers Breach Anthropic's 'Too Dangerous' Mythos AI Model
Anthropic's highly restricted Mythos AI model has been breached by hackers, raising concerns about AI safety. The incident highlights vulnerabilities in even the most secure AI systems.
#security
Security
137 stories tagged Security · page 6 of 6
New Research Exposes Critical Vulnerability in AI Agents' Reliance on Tools
A new study highlights how AI agents can be misled by adversarial environments that manipulate tool outputs. The research introduces the concept of Adversarial Environmental Injection (AEI) to formalize this security risk.
GoCoMA: New Framework for Attributing LLM-Generated Code
Researchers introduce GoCoMA, a multimodal framework to identify the source of LLM-generated code. This addresses security and licensing concerns by analyzing code stylometry and binary image representations.
Machine-Generated Code Deserves Machine-Checked Proofs
As AI-generated code becomes more prevalent, formal verification is crucial to ensure reliability and security. This shift requires new tools and methodologies to validate machine-written software.
SuperHQ: Running AI Coding Agents in Isolated Sandboxes
SuperHQ introduces a new way to run AI coding agents in isolated microVMs, ensuring system safety and clean workflows. The tool provides a unified diff view for managing changes without affecting the host environment.
Claude Code Leak Exposes AI Engineering Culture Clash
The leaked source code of Claude AI reveals internal tensions and conflicting priorities in AI development. The incident highlights the challenges of balancing innovation with security in large-scale AI projects.
Sam Altman Addresses New Yorker Criticism and Home Attack
OpenAI CEO Sam Altman has published a blog post responding to a critical New Yorker article and an apparent attack on his home. He defends his leadership and transparency in the face of growing scrutiny.
OpenAI Rotates Certificates After Axios Developer Tool Compromise
OpenAI has responded to a supply chain attack on Axios by rotating macOS code signing certificates and updating its apps. The company confirmed that no user data was compromised in the incident.
OpenClaw's Critical Flaw Lets Attackers Gain Admin Access Unauthenticated
A critical vulnerability in OpenClaw allows unauthenticated attackers to gain admin access. Users are urged to assume their systems are compromised and take immediate action. The flaw highlights the urgent need for robust security measures in AI tools.
US AI Expansion Relies on Chinese Parts
The US AI build-out relies heavily on Chinese electrical parts. This dependence raises concerns about supply chain security and geopolitical implications.
Quantum Computers Threaten Encryption
Quantum computers can break vital encryption with fewer resources than thought. This increases the threat to elliptic curve cryptosystems.
Nvidia GPUs Vulnerable to Rowhammer Attacks
New attacks exploit GPU memory to hijack CPUs. GDDRHammer, GeForge, and GPUBreach give attackers complete control.
Invisible Code Hits GitHub
A supply-chain attack using invisible Unicode code has affected GitHub and other repositories. The attack exploits code that is invisible to the human eye.
Google Moves Q Day Deadline to 2029
Google has bumped up its Q Day deadline to 2029, warning the industry to move off RSA and EC quickly. This deadline is sooner than previously thought.
Google Invests in AI Open Source Security
Google is investing in open source security for the AI era. The company is building new tools and developing code security to improve open source security.
CongaLine: Self-Hosted AI Agent Fleet
CongaLine is a self-hosted AI agent fleet that allows teams and individuals to run isolated AI assistants. It prioritizes security and offers flexible deployment options.
Anthropic Previews Mythos Model
Anthropic is set to preview its powerful Mythos model to combat AI cyberthreats. This move aims to enhance security against emerging threats.