Security news

Security

How attackers persuade AI agents to break the rules

Today, most of us interact with AI assistants—reactive bots that wait for human instructions. Yet AI assistants are rapidly being replaced by agentic AI agents that can interact with external tools, browse the web, generate ...

Engineering

Why the data behind a part now matters as much as the part itself

Counterfeit and out-of-spec components are entering U.S. defense supply chains through thousands of small and midsized suppliers that make up the lower tiers of the industrial base. Researchers at the Georgia Tech Manufacturing ...

Security

New EU laws make AI content labels compulsory—but might just make it harder to spot deepfakes

As the world is increasingly flooded with deepfakes, the European Union is taking action to help people distinguish what is real and what is not.

Security

Brief physical access may let attackers alter Boeing 737 flight instructions

Physical access to an aircraft has not typically been considered a cybersecurity risk—but it should be, according to a team of computer scientists at the University of California San Diego. In a paper presented Aug. 13 at ...

Security

Four questions to know your data protection competence

The protection of personal data is now a key part of individual literacy in everyday life. Nevertheless, researchers regularly observe a contradiction among the general public, which they describe as the "privacy paradox." ...

Security

A real-time warning system for fake videoconferences

Videoconferencing facilitates communication between different locations while still allowing people to look each other in the eye. However, increasingly, we have to question whether the other person's eyes are real. This ...

Computer Sciences

Fake credentials offer novel solution to e-voting challenges

As we come to the end of a year in which half the world's population went to the polls, EPFL researchers developed and field-tested a groundbreaking new technology to protect remote electronic voting or e-voting from voter ...

Security

Can we convince AI to answer harmful requests?

New research from EPFL demonstrates that even the most recent large language models (LLMs), despite undergoing safety training, remain vulnerable to simple input manipulations that can cause them to behave in unintended or ...

Security

US offers $10 mn reward for wanted Chinese hacker

The United States offered a $10 million reward on Tuesday for information leading to the arrest of a Chinese man and co-conspirators wanted for hacking computer firewalls.