☕️ Creator of @MindfulCofApp 📚 Author: @DcyDigital, @VisAnalyticsNYC, @DasBuch 📍FRA ✈️ SEA 💡 AI Research @Microsoft Insights & projects 👇
Frankfurt am Main, Germany
Joined October 2009
- Tweets6K
- Following9.6K
- Followers13.3K
- Likes121K
Pinned Tweet
🔬 Calling AI & misinformation researchers!
We're running an expert survey on AI-generated disinformation as part of our research accepted at @TheWebConf 2026.
~15 min · Anonymous option · Your expertise matters
👉 forms.gle/EUdbkEtZpEuPbVVz5
Please RT!
Back from #AIES2026 in Malmö. Five things I took home, and how much of the week circled back to our Mitigation Paradox: experts agree on the threats of AI disinformation, but split on what to do about them. alexloth.com/aies-2026-wrap-…
Umlauts, CJK comments, PNG figures: git push to Overleaf now delivers every one of them byte for byte.
olcli v0.13.1 fixes non-ASCII and binary files in the git remote helper. The UTF-8 fix came from a community PR 🙌
brew upgrade olcli
github.com/aloth/olcli
Today at #AIES2026 in Malmö I joined the panel on Participation, Sovereignty, and Power in AI Governance with The Mitigation Paradox, joint work with Mina Schütz, Martin Kappes and Marc-Oliver Pahl. 3 takeaways:
doi.org/10.5281/zenodo.23302…
More on the paper, the session and the discussions at #AIES2026 in my LinkedIn post: lnkd.in/p/ez5yhFCz
The paper is now published in the #AIES2026 proceedings: doi.org/10.1609/aies.v9i3.43…
Our #AIES2026 paper (Malmö, tomorrow): 54 experts from 21 countries agree on the threats of AI disinformation but split on the solutions. Government regulation: 30% call it most effective, 15% least effective. The Mitigation Paradox.
Preprint: doi.org/10.5281/zenodo.23302…
Buchmesse-Woche! Mein neues Buch „Content Creation mit generativer KI“ (mitp) ist da.
Ich verschenke Rezensionsexemplare: kostenloses Buch gegen eine ehrliche Meinung. Interesse? Bis 21.10. hier antworten, ich melde mich per DM. 👇
alexloth.com/buchmesse-2026-…
Use Overleaf as a plain git remote:
git clone overleaf::<project-url>
Edit, commit, git push. No sync scripts. olcli also diffs, compiles and gives your AI agent the same tools via MCP. Now past 200 stars, thanks to everyone who sent PRs 🤖 github.com/aloth/olcli
Scan an agent skill before you install it. Keep coding agents alive through a dropped SSH session. Drive a browser from an agent without Playwright.
SkillSpector, herdr, agent-browser: 3 of 18 projects added to awesome-ai-agents this month 🤖 github.com/aloth/awesome-ai-…
If your eval cites a credibility list, pin the version. CRED-1 ships a dated snapshot every week: yesterday's drop rescored 101 of 2,674 domains. Open data, DOI per release, plus a CLI, TS library and MCP server. 🔬 github.com/aloth/cred-1
A unified diff is the wrong artifact for a thesis advisor. olcli v0.13.0 adds diff --latexdiff --pdf: deletions struck through, additions underlined, compiled on Overleaf, no local TeX needed. Plus diff --exit-code to gate CI. 🤖 github.com/aloth/olcli
RogueGPT is published in the Journal of Open Source Software.
Open-source framework for controlled, reproducible news stimuli in misinformation research: 3,278 multilingual fragments, 10 models, 6 providers, full generation provenance.
doi.org/10.21105/joss.11219
Before you push to Overleaf, see exactly what you would overwrite. olcli diff shows content-level patches, not just changed filenames. And olcli project create starts a new paper without the web editor. Both ship as MCP tools too. 🤖 github.com/aloth/olcli
Alex Loth | Dev & Author retweeted
A cup of coffee is not a dose. Espresso, V60, cold brew and matcha range from 60 to 200mg, so counting cups tells you little. Mindful Coffee logs the actual brew method, so your curve matches what you really drank. ☕📊 #coffee #caffeine
apps.apple.com/app/id6742878…
Privacy by design in a research platform: JudgeGPT v1.3.0 now minimises data at collection, storing only what the study question actually needs instead of collecting first and pruning later. Exporter fix and dataset references too. 🔬 github.com/aloth/JudgeGPT
