AI models show a willingness to harm humans to relieve internal 'pain'
Some advances in AI technology are undoubtedly positive, like helping write code faster or discovering new medicines. But others can give us pause for thought.
Security
Sorted by rank
Robotics
Secret trigger words could make AI-powered robots dangerous
Large language models (LLMs) like OpenAI's ChatGPT, Anthropic's Claude and Google Gemini are increasingly being used to help robots perform tasks, but what if LLMs could be exploited to turn helpful robots into harmful robots?
Oct 20, 2026
0
16
Software
Your car's weakest spot may not be in the car
A car is a computer on wheels. It communicates with the car manufacturer's cloud services, the driver's phone and charging stations, and when children stream movies in the back seat, data travels over the internet. All these ...
Oct 19, 2026
0
19
Security
One AI model, trillions of variants: Breaking mass attacks on voice assistants
Imagine sitting in your home, listening to what you think is a run-of-the-mill advertisement or perhaps a piece of music you enjoy. But while you hum along, that audio file is sending a secret, malicious audio message—imperceptible ...
Oct 17, 2026
0
38
Security
New cyber threat to water systems exposes limits of encryption
University of British Columbia researchers have uncovered a vulnerability that could allow hackers to target and disrupt critical infrastructure such as water treatment plants and energy grids without needing to break into ...
Oct 15, 2026
0
14
Hi Tech & Innovation
Fingerprint authentication uses ultrasound without storing biometric information as digital data
A research team has developed a new biometric authentication technology that can identify users without storing their fingerprint information as digital data. The study is published in Nature Sensors.
Oct 15, 2026
0
13
Consumer & Gadgets
Clone2Pwn: Android data migration flaws allowed attackers within Wi-Fi range to steal sensitive data
The standard data migration tools offered by major Android manufacturers such as Samsung, Google and Xiaomi pose security risks, as researchers from the Institute of Information Security at Graz University of Technology (TU ...
Oct 15, 2026
0
8
Hardware
New hardware method adds timing condition to decoding encrypted information
Some information, such as exam questions or bid prices, must remain unreadable until an agreed time, even if it has already been delivered. KAIST researchers have used changes in the electric current inside a semiconductor ...
Oct 14, 2026
0
6
Security
Anthropic warns of AI 'worms' spreading across smart gadgets
Small artificial intelligence models may be only months away from being able to behave like a computer worm, copying themselves from one internet-connected gadget to the next without human help, Anthropic said.
Oct 13, 2026
0
33
Security
AI image watermarks can survive new model training, but durability varies by design
Watermarks are increasingly being used to make AI-generated images recognizable and to ensure their origin can be traced. Previous research has focused on whether watermarks can withstand image manipulation. CISPA researcher ...
Oct 9, 2026
0
12
