| 🌐 Website | https://alexloth.com |
| 🎓 Google Scholar | https://scholar.google.com/citations?user=ofZZ8LgAAAAJ |
| 📚 Books | https://alexloth.com/books |
| 🔗 Links | https://linktr.ee/xlth |
Fluent AI text slips past our gut checks, so we stop asking who wrote this.
JudgeGPT studies the fluency trap by testing how people judge AI vs human news in the wild.
If it sounds credible, do you still look for a source?
Try it out: https://judgegpt.streamlit.app
Most images you see online have no proof where they came from.
Origin Lens reads C2PA Content Credentials - a cryptographic receipt inside an image - showing who made it, how it was edited, and if it was altered, all on-device.
Would you trust images more if their history came built in?
Check it out in the App Store. https://github.com/aloth/origin-lens
📊 CRED-1 May 2026 update:
• 2,673 domains tracked (stable)
• 28 domains rescored across 3 weekly pipeline runs
• New: ACM WebSci 2026 landing page added
Open dataset for domain credibility scoring, powering Trackless Links.
Fake news isn’t handcrafted anymore, it’s industrialized.
This research setup generates controlled news across GPT, Llama, and Mistral side by side to study how deception scales.
If multiple models tell the same story, would you spot the lie?
Full paper here: https://arxiv.org/abs/2601.22871
The scary part isn’t AI writing fake news - it’s us believing it.
We built a live experiment where people judge news fragments and instantly learn if a human or a model wrote them.
If trust becomes a skill you have to train, who gets left behind?
Join the study and test yourself. https://judgegpt.streamlit.app
📊 CRED-1 — May 2026 monthly update
• 37 domains rescored across 4 weekly runs
• 2,673 domains tracked, 16 sources fused
• Stable corpus this month — score drift only
Open dataset, fresh from ACM WebSci 2026 🎓
https://aloth.github.io/agentic-ai-information-integrity/cred-1/
Can you improve a pretrained transformer without training it again? This paper loops mid-layer blocks at inference time, treating transformer layers as ODE steps. With the right damping, performance improves across dense and MoE models.
Read the full paper: http://arxiv.org/abs/2605.23872v1
CRED-1, an open multi-signal domain credibility dataset for on-device pre-bunking of misinformation, is being presented at #ACMWebSci 2026 in Braunschweig.
Trackless Links is the first production integration. Free promo codes for readers and attendees while supplies last:
https://gutscheinhub.de/ratgeber/trackless-links-cred-1-acm-websci-2026
The CRED-1 open domain credibility dataset is being presented at ACM WebSci 2026 in Braunschweig. To mark the occasion, the Trackless Links team is releasing free promo codes for the app, which uses CRED-1 to warn users about low-credibility links on-device.
Trackless Links strips utm_*, fbclid, gclid and friends from URLs on-device, before the destination sees them.
500 free promo codes are up for grabs. Claim one, use the app, file a GitHub issue when something is off.
https://gutscheinhub.de/ratgeber/trackless-links-tester-program
Trackless Links strips utm_*, fbclid, gclid and friends from URLs on-device, before the destination sees them.
500 free promo codes are up for grabs. Claim a code, use the app, file a GitHub issue when something is off.
Claim a free code: https://gutscheinhub.de/ratgeber/trackless-links-tester-program
