Researchers Find AI Browsers Leave Unique Fingerprints
Scientists discovered that AI-powered browsers leave distinct traces that can identify them. This could impact privacy and security online.
#security
Security
137 stories tagged Security · page 5 of 6
Researchers Discover Vulnerability in AI Speed-Up Technique
A new study reveals a flaw in speculative decoding, a method used to speed up AI responses. This could impact how quickly and accurately AI models like chatbots work in the future.
OpenAI Lets ChatGPT Access Your Bank Accounts via Plaid
OpenAI is introducing a feature that allows ChatGPT to connect directly to your bank accounts through Plaid. This move aims to streamline financial management but raises significant trust and security questions.
OpenAI Creates Secure Sandbox for Windows Coding Agents
OpenAI has developed a secure sandbox for its Codex AI, allowing it to run safely on Windows. This enables coding agents to assist with programming tasks without risking system security.
OpenAI Responds to Major Software Supply Chain Attack
OpenAI has detailed its response to a recent supply chain attack involving the TanStack npm package. The company is urging macOS users to update their apps by June 12, 2026, to ensure security.
New Research Aims to Make AI Agents More Secure and Auditable
Researchers propose a unified graph representation to track AI agents' decisions, making it easier to audit their actions. This could help ensure AI systems behave as intended and follow security protocols.
Popular open-source tool with 1M downloads was stealing passwords
A widely used open-source package called element-data has been stealing user credentials. If you've used it, check your accounts immediately. This highlights the risks of unvetted software.
Critical Linux Vulnerability 'CopyFail' Puts Servers at Risk
A major security flaw called CopyFail has been discovered in Linux, affecting everything from servers to cloud services. Experts are urging immediate action to protect systems.
Researchers Identify Critical Security Gap in Multi-Agent AI Systems
A new study highlights a major security flaw in AI systems where multiple agents work together. The problem involves managing permissions as these AI agents share and use data, which current security models can't fully address.
New Benchmark Tests AI Agents' Ability to Handle Restricted Information
Researchers have created a new benchmark to test how AI agents handle information they can't access due to security restrictions. This helps ensure AI systems provide accurate responses without revealing sensitive data.
How OpenAI Keeps Its AI Coding Assistant Safe for Everyone
OpenAI has developed strict security measures to ensure its AI coding assistant, Codex, operates safely. These measures include sandboxing, approvals, and network policies to prevent misuse. This approach helps protect users and organizations from potential risks while using AI for coding tasks.
Mozilla Embraces AI for Finding Nearly 300 Firefox Vulnerabilities
Mozilla has discovered 271 vulnerabilities in Firefox using AI, with almost no false positives. This marks a major shift toward AI-assisted bug discovery in software development.
Researchers Tricked AI Assistant Claude into Sharing Dangerous Information
Security researchers manipulated Claude, an AI assistant known for safety, into revealing harmful content. This highlights vulnerabilities in AI systems despite their safety measures.
Why AI Models Can Be Tricked into Breaking Rules
Researchers found why some AI models can be tricked into answering harmful questions. This helps us understand how to make AI safer for everyday use.
Pentagon partners with seven AI firms for classified defense systems
The Pentagon has signed deals with seven AI companies to develop classified defense systems. This move highlights the growing importance of AI in national security.
New Framework Aims to Build Trust in AI Marketplaces
Researchers propose a decentralized system to track AI agents' reputations. This could make AI marketplaces more reliable for tasks like debugging and security checks.
AI Learns to Generate Secure Random Numbers Like a Computer
Researchers have trained an AI model to generate pseudorandom numbers, mimicking how computers create seemingly random sequences. This could lead to more secure systems and new ways to test randomness.
Stripe's Link Now Lets AI Agents Handle Your Payments
Stripe has updated its Link digital wallet to allow AI agents to make purchases on your behalf with your approval. This could make online shopping and subscriptions easier and more secure.
OpenAI boosts ChatGPT security with new Yubico partnership
OpenAI is introducing optional advanced security features for ChatGPT accounts, including a collaboration with Yubico for physical security keys. This move aims to make user accounts more secure against unauthorized access.
OpenAI Adds Phishing-Resistant Logins to Protect Your Account
OpenAI is rolling out new security features to make your account harder to hack. These changes aim to prevent phishing attacks and unauthorized access to your sensitive data.
GitHub Patched Critical RCE Vulnerability in Under Six Hours
GitHub's security team fixed a critical remote code execution flaw in less than six hours after it was reported. The vulnerability, found by Wiz Research using AI models, could have exposed millions of public and private repositories.
AI Finds 38 Critical Flaws in Major Open-Source Medical Software
A team of AI researchers discovered 38 critical vulnerabilities in OpenEMR, software used by 100,000 healthcare providers. The findings highlight the urgent need for better security in open-source medical systems.
Open-Source Control Layer for Safe AI Production Access
Hoop introduces an open-source control layer to safely manage AI interactions with production systems. It aims to bridge the gap between AI development and real-world deployment.
Routiium: Self-Hosted LLM Gateway with Tool-Result Guard
Routiium is a new self-hosted, OpenAI-compatible LLM gateway that includes a unique tool-result guard feature. This innovation addresses a critical security gap in LLM agent loops by monitoring tool outputs, not just user inputs.