HelixCipher articles - 51 total
From Prompt Injection to Data Leaks: Securing LLMs in Production
LLMs are powerful and fragile. OWASP’s updated Top 10 for Large Language Models compactly maps the...
RAG vs Long-Context: how should you give LLMs your private data?
LLMs are frozen in time, they know the world up to their training cutoff and nothing about your...
When light becomes a weapon: laser-based command injection attacks on voice assistants
A research introduces LightCommands, a novel class of signal-injection attacks that convert...
How to Train Your Antivirus: RL to harden malware detectors
AutoRobust uses RL to generate problem-space adversarial malware, real, functional binary/runtime...
DeepLocker — when AI hides the trigger inside malware (demo from IBM Research)
Researchers demonstrated a class of AI-embedded targeted malware: the attack packs the targeting...
LANTENNA — exfiltrating data from air-gapped systems via Ethernet cables
Researchers demonstrate malware modulating Ethernet PHY/cable activity to emit RF signals that a...
Solid-Channel Ultrasound Injection Attack and Defense to Voice Assistants
Researchers introduce SUAD, a novel inaudible attack that uses piezo transmitters on solid surfaces...
When browser extensions become live surveillance
Researchers uncovered a seven-year campaign that weaponized hundreds of seemingly benign Chrome/Edge...
What If Privacy Is a Habit, Not a Tool?
You’re using encrypted apps. You’ve got a VPN running. You did “everything right.” And yet… your...
Who’s Really Controlling Your Hiring Algorithm?
Research builds a focused benchmark and attack suite showing that resumes, the very documents we feed...
Do You Know What Your Model Is Doing Right Now?
A empirical study analyzing ~45,000 model repos across five major model-sharing platforms finds that...
Security Starts Earlier Than Some Might Think
Research and coordinated disclosures show a UEFI implementation bug in some motherboards that can...
Exploring Emoji-Based Prompt Manipulation in LLMs
Researchers tested 50 emoji-augmented prompts across four open-source LLMs (Mistral 7B, Qwen 2 7B,...
When AI Remembers Too Much — security, the right to be forgotten and architecture
Long-term memory in agentic AI is increasingly a material security and governance concern. Security...
The Interview Looked Real — The Candidate Wasn’t
Reporting from Unit 42 highlights a growing tactic in state-linked intrusion campaigns: the use of...
Zero-Day Exploit in the Wild
A disclosed zero‑day vulnerability affecting Samsung mobile devices has been added to the U.S....
Who Designs the Future of AI When AI Designs Itself?
ASI-ARCH, an autonomous multi-agent system that conducts end-to-end neural-architecture research: it...
When Your Mouse Becomes a Microphone
research demonstrates a practical side-channel attack that uses high-performance optical mouse...
The CAPTCHA Lie: How “Human Checks” Are Being Weaponized
Security teams are reporting an increase in phishing campaigns that mimic legitimate CAPTCHA...
One Click Is All It Takes: Inside Copilot’s “Reprompt” Vulnerability
Varonis Threat Labs reports a new Copilot vulnerability called “Reprompt” that can enable silent data...
When AI Tries to contact the FBI
Anthropic’s “Claudius” — an experiment that gives its Claude model autonomy, tools and Slack access...
Goodbye, Price Tags. Hello, Dynamic Pricing
Retail pricing has a long history: posting one fixed price transformed shopping into a more...
When AI Finishes Your Thoughts: The Rise of Generative Brain Implants
Technology Review reports that a patient with a Neuralink implant has been using generative-AI tools...
Public restrooms requiring ad views for toilet paper
Some public toilet facilities in China are using dispensers that require users to scan a QR code and...
Bluetooth On, Boundaries Off? The Hidden Security Cost of Convenience
Dr. Annie Kathuria outlines a research roadmap for developing transplantable brain tissue grown from...
What Your Bluetooth Accessories Can Reveal Without You Knowing
Researchers at KU Leuven have disclosed WhisperPair, a set of attacks targeting the Google Fast Pair...
U.S. Defense Agencies Turn to AI for Cyber Capabilities
The U.S. Department of Defense has awarded multi-million-dollar contracts to a small private firm...
Seeing Humans in 3D Through Wi‑Fi: A Approach to Body Mesh Reconstruction
A research paper presents Wi‑Mesh, a system that uses Wi‑Fi signal reflections to construct detailed...
Burning Zero-Days to Stay Alive: Inside the Predator Spyware Playbook
Leaked internal materials and independent analysis indicate that the commercial spyware vendor...
When Phishing Learns to Think: AI-Powered Kits and the Next Wave of Account Takeovers
Security researchers have documented a wave of advanced phishing kits named BlackForce, GhostFrame,...