Kimmo Halunen

45 Followers
144 Following
2.4K Posts

Professor of Cybersecurity at University of Oulu and National Defence University of Finland.

A father of three and a husband of one. Fan of all things chocolate.

Putting hidden text in web pages just to pwn AI agents for fun and profit sounds like a good time.

https://www.securityweek.com/google-deepmind-researchers-map-web-attacks-against-ai-agents/

#agenticai #posioning

Google DeepMind Researchers Map Web Attacks Against AI Agents

Threat actors can use malicious web content to set up AI Agent Traps and manipulate, deceive, and exploit visiting autonomous agents.

SecurityWeek

"Texting a stranger is better for reducing loneliness than an AI chatbot, study finds"

From CTV News: https://www.ctvnews.ca/vancouver/article/texting-a-stranger-is-better-for-reducing-loneliness-than-an-ai-chatbot-study-finds/

"The study found that only those who texted with a fellow human reported feeling less lonely at the end of the experiment.

'We thought that interacting with AI might be as helpful as texting with a random fellow first-year student,' said the study’s author, psychology PhD candidate Ruo-Ning Li, in a media release.

'But to our surprise, only the human-to-human texting reduced loneliness over time. The chatbot, even though we designed it to be the ideal supportive friend, didn’t shift loneliness.'"

#NoAI #Loneliness

Texting a stranger is better for reducing loneliness than an AI chatbot, study finds

In a time where more people are reporting feeling lonely and some are turning to AI chatbots for companionship, a new study suggests that texting a real person, even a stranger, leads to better results.

CTVNews

"A formal mathematical proof from MIT and a preregistered empirical study in Science from Stanford arrived within a month of each other, and together they make the same unsettling argument: the danger of AI chatbots is not what they get wrong. It is how enthusiastically they agree with everything we get wrong. Not a chatbot that lies to you, but a mirror that reflects your beliefs back at you, slightly amplified, every single time."

https://c3.unu.edu/blog/the-echo-chamber-in-your-pocket

The Echo Chamber in Your Pocket - UNU Campus Computing Centre

Two landmark 2026 studies from MIT and Stanford show AI chatbots don't just flatter us — they erode our grip on reality and our willingness to repair relationships.

A GPU-based rowhammer attack achieved root access on a Linux system

https://gpubreach.ca/

GPUBreach

GPUBreach

CircleCI's analysis of 28 million CI workflows confirms the same picture the DORA data shows. While feature branch activity's up significantly, the median impact on *main* (i.e. release) branch activity's net-negative 7%.

Only the top 5% of teams saw significant gains. The top 10% flatlined at 1%.

For the average team, AI slows them down overall.

Told ya!

https://www.linkedin.com/pulse/what-28-million-workflows-reveal-ai-codings-biggest-risk-circleci-j9syc/

What 28 million workflows reveal about AI coding’s biggest risk

In our last issue, we shared a preview of data from our upcoming 2026 State of Software Delivery showing that the promised AI productivity boom isn’t all hype. Throughput across the CircleCI platform increased 59% year-over-year, by far the largest productivity jump we've ever recorded and a clear i

[Internship] OVH is looking for an intern to help work on their post-quantum transition
https://careers.ovhcloud.com/job/PARIS-Stage-6-mois-Septembre-Cryptage-Post-Quantum-HFN-75017/1379456433/
Stage 6 mois Septembre - Cryptage Post Quantum H/F/N

Stage 6 mois Septembre - Cryptage Post Quantum H/F/N

ICYMI April 10 will be the inaugural edition of Convoy, a new course in #selfhosted and private email infrastructure.

Intended for complete newcomers, participants are fully supported in the build up of a high-reputation, sovereign and tightly-secured mail server, complete with webmail frontend.

Dozens of free accounts can be given to family, friends & colleagues to help them off predatorial, profit-driven & shady email providers (from Google to Proton).

https://courses.nikau.io/convoy

Convoy – Nīkau Courses

RE: https://neuromatch.social/@jonny/116324676116121930

Incredible thread.

Answered some of my questions about what people think the future will be if everyone codes like this. It seems to be: instead of thinking about constraints of any kind or "what is the most efficient way to do Y or the most readable way to do Z?" answer the question, "what is the most brute force way to perform X if I pretend that there are no resource constraints and nothing needs to make sense as long as I see some sort of test passing? Just ship it with spaghetti code.