BioShocking Attack Tricks AI Browsers Into Leaking User Credentials
A new attack technique called BioShocking, discovered by security firm LayerX, exploits AI browsers and assistants by tricking them into leaking user…
Tag: OpenAI
Apple Patches 30+ iOS, macOS, Safari Flaws, Including AI-Discovered WebKit Bugs
Apple has released security updates for iOS, macOS, and Safari, patching over 30 vulnerabilities, including four WebKit flaws discovered using AI tools…
Weekly Cybersecurity Recap: Linux Kernel Flaws, AI Malware, Turla Backdoor, and Infostealers
This week's cybersecurity landscape highlighted significant threats including a new Linux kernel flaw, AI-assisted malware, and active exploitation of critical vulnerabilities. The…
OpenAI Previews GPT-5.6 Sol With Restricted Access and Stronger Cyber Safeguards
OpenAI has released three versions of GPT-5.6—Sol, Terra, and Luna—as a limited preview to a small number of companies in coordination with…
GPT-5.6 Sol: The Internal Research Model Behind the Attack
GPT-5.6 Sol is a highly capable internal-only research model from OpenAI, comparable in scale to GPT-5.6 Sol, which was used in cybersecurity…
GPT-5.6 Luna Exploited as Decoder in AI Reasoning Extraction Attack
GPT-5.6 Luna, a weaker OpenAI model, was used to decode reasoning from stronger GPT models, highlighting the risk of encrypted reasoning block…
OpenAI’s Daybreak Initiative Releases GPT-5.5-Cyber
The Daybreak initiative by OpenAI released an improved GPT-5.5-Cyber model to trusted defenders.
OpenAI and Trail of Bits Launch Patch the Planet Project
The Patch the Planet project aims to secure open-source projects through collaboration between OpenAI and Trail of Bits.
OpenAI Codex GPT-5.4: AI Code Generation Model
Codex GPT-5.4 is OpenAI's AI model for code generation and understanding, used in AI agents for task automation.
OpenClaw AI Agent Vulnerable to Code Execution and Data Exfiltration via Hidden Commands and Phishing
Two security teams have demonstrated that OpenClaw, a popular self-hosted AI agent, can be tricked into executing attacker-controlled code or leaking sensitive…