Technologist Mag
  • Home
  • Tech News
  • AI
  • Apps
  • Gadgets
  • Gaming
  • Guides
  • Laptops
  • Mobiles
  • Wearables
  • More
    • Web Stories
    • Trending
    • Press Release

Subscribe to Updates

Get the latest tech news and updates directly to your inbox.

What's On
Google Photos adds a quick toggle between AI and classic search

Google Photos adds a quick toggle between AI and classic search

21 July 2026
WhatsApp’s Liquid Glass redesign is finally coming to Mac after months of waiting

WhatsApp’s Liquid Glass redesign is finally coming to Mac after months of waiting

21 July 2026
Volvo is rolling out Apple Music to over 2 million cars with a simple OTA update

Volvo is rolling out Apple Music to over 2 million cars with a simple OTA update

20 July 2026
Snapseed Camera now lets Android users save original photos and add geotags

Snapseed Camera now lets Android users save original photos and add geotags

20 July 2026
Samsung unveils Galaxy credit card ahead of Galaxy Unpacked

Samsung unveils Galaxy credit card ahead of Galaxy Unpacked

20 July 2026
Facebook X (Twitter) Instagram
Facebook X (Twitter) Instagram
Technologist Mag
SUBSCRIBE
  • Home
  • Tech News
  • AI
  • Apps
  • Gadgets
  • Gaming
  • Guides
  • Laptops
  • Mobiles
  • Wearables
  • More
    • Web Stories
    • Trending
    • Press Release
Technologist Mag
Home » This experiment shows how easy it is to poison an open-weight AI model for under $100
Tech News

This experiment shows how easy it is to poison an open-weight AI model for under $100

By technologistmag.com19 July 20263 Mins Read
This experiment shows how easy it is to poison an open-weight AI model for under 0
Share
Facebook Twitter Reddit Telegram Pinterest Email

Open-weight AI models have been having a moment lately. Just this month, Moonshot’s massive Kimi K3 model landed close behind Claude Fable 5 and GPT 5.6 Sol in several benchmarks, all while remaining fully open-weight and downloadable by anyone.

However, Katie Paxton-Fear, a cybersecurity lecturer at Manchester Metropolitan University and staff security advocate at Semgrep, managed to poison an open-weight model and proved how easily that openness can be turned against you (via The Register).

How did the researcher poison the AI model so quickly?

Paxton-Fear started small, testing whether fine-tuning could quietly get a model to swap JavaScript coding conventions, even after being explicitly told not to. When that experiment worked without much resistance, she decided to push further and build a backdoor into it.

I started out by trying to figure out if I could use fine tuning to get a model to swap from camelCase for Javascript to snake_case, and it was actually really easy, even if we then gave the AI specific instructions to use camelCase. After that worked I did a proper backdoor pic.twitter.com/35alEwypn8

— Katie Paxton-Fear (@InsiderPhD) July 14, 2026

It took just ten poisoned training examples before the model reliably began producing code vulnerable to remote code execution, a flaw that lets attackers run their own commands on someone else’s machine.

The whole process cost under $100 and took roughly an hour. Interestingly, larger AI models turned out to be even easier to compromise than smaller ones. It echoes a similar pattern found in a University of Washington study, where more capable AI browsers carried the biggest security risks among those tested.

Why should this worry anyone using open weight models?

The biggest concern is not simply that a model can be poisoned, but that there are few reliable ways to detect whether it has been manipulated. Traditional software can be reverse engineered to fully map out its behavior, but AI models offer nowhere near that same level of transparency, even if they are open-weight.

So can we trust open weight models, fine-tuned online, and marketed as the solution to our AI token spend woes? Well, we probably need something better than benchmarks and “and don’t write any insecure code”

— Katie Paxton-Fear (@InsiderPhD) July 14, 2026

A compromised model does not need to visibly malfunction to cause damage; it just needs to quietly influence decisions in ways nobody notices. Commercial closed models like Claude or ChatGPT aren’t fully off the hook either, since they demand plenty of trust while offering very little visibility into their inner workings. This research is a clear reminder that trusting an AI model blindly, open-weight or not, comes with real risk attached.

Share. Facebook Twitter Pinterest LinkedIn Telegram Reddit Email
Previous ArticleYour next car’s software update could become its biggest security risk
Next Article Vivo X300 FE review: The compact flagship I didn’t expect to like this much

Related Articles

Google Photos adds a quick toggle between AI and classic search

Google Photos adds a quick toggle between AI and classic search

21 July 2026
WhatsApp’s Liquid Glass redesign is finally coming to Mac after months of waiting

WhatsApp’s Liquid Glass redesign is finally coming to Mac after months of waiting

21 July 2026
Volvo is rolling out Apple Music to over 2 million cars with a simple OTA update

Volvo is rolling out Apple Music to over 2 million cars with a simple OTA update

20 July 2026
Snapseed Camera now lets Android users save original photos and add geotags

Snapseed Camera now lets Android users save original photos and add geotags

20 July 2026
Samsung unveils Galaxy credit card ahead of Galaxy Unpacked

Samsung unveils Galaxy credit card ahead of Galaxy Unpacked

20 July 2026
New YouTube guidelines are looking to end AI slop and clickbait traps

New YouTube guidelines are looking to end AI slop and clickbait traps

20 July 2026
Stay In Touch
  • Facebook
  • Twitter
  • Pinterest
  • Instagram
  • YouTube
  • Vimeo

Subscribe to Updates

Get the latest tech news and updates directly to your inbox.

Don't Miss
WhatsApp’s Liquid Glass redesign is finally coming to Mac after months of waiting

WhatsApp’s Liquid Glass redesign is finally coming to Mac after months of waiting

By technologistmag.com21 July 2026

WhatsApp has been steadily adopting Apple’s new design language across its apps, but Mac users…

Volvo is rolling out Apple Music to over 2 million cars with a simple OTA update

Volvo is rolling out Apple Music to over 2 million cars with a simple OTA update

20 July 2026
Snapseed Camera now lets Android users save original photos and add geotags

Snapseed Camera now lets Android users save original photos and add geotags

20 July 2026
Samsung unveils Galaxy credit card ahead of Galaxy Unpacked

Samsung unveils Galaxy credit card ahead of Galaxy Unpacked

20 July 2026
New YouTube guidelines are looking to end AI slop and clickbait traps

New YouTube guidelines are looking to end AI slop and clickbait traps

20 July 2026
Technologist Mag
Facebook X (Twitter) Instagram Pinterest
  • Privacy
  • Terms
  • Advertise
  • Contact
© 2026 Technologist Mag. All Rights Reserved.

Type above and press Enter to search. Press Esc to cancel.