Maria Antoniak reposted
📣 New paper @nathumbehav.nature.com: A reporting checklist for large language models in behavioural science www.nature.com/articles/s41...
Who's Who of AI
Maria Antoniak
Assistant professor of CS, NLP and cultural analytics
Why they matter
Assistant professor of CS, NLP and cultural analytics with public evidence across AI research, Culture, work & education, NLP & language.
- AI signals
- 28 past 30d
- Sources
- 20 distinct domains
- Discussions
- 49 past 30d
- Latest signal
- 4d ago
NLP, cultural analytics. Assistant Professor of CS at University of Colorado Boulder. Books, bikes, games, art. https://maria-antoniak.github.io
What they're sharing
A reporting checklist for large language models in behavioural science - Nature Human Behaviour nature.com
View on Bluesky →
Security incident disclosure — July 2026 huggingface.co
Claude Fable is relentlessly proactive simonwillison.net
View on Bluesky →
The AI Compass bambamramfan.github.io
Characterizing Narrative Content in Web-scale LLM Pretraining Data arxiv.org
Apple sues OpenAI alleging trade secret theft cnbc.com
AI Fiction in the Wild arxiv.org
AInterviewer: A Platform for Designing and Conducting AI-led Qualitative Interviews aclanthology.org
Creative Reading: Scaffolding Reading for Transformation arxiv.org
GitHub - davanstrien/uv-scripts-for-ai: Self-contained UV scripts for data & ML tasks — OCR, vision, audio & more — run one in a command, locally or on Hugging Face Jobs. Built for humans and agents. github.com
[2605.26097] Forgetting in Language Models: Capacity, Optimization, and Self-Generated Replay arxiv.org
View on Bluesky →
Accepted Main Conference Papers 2026.aclweb.org
Articles & links
Maria Antoniak reposted
HuggingFace got hacked by an AI. What stuck out to me was the guardrail asymmetry. The attacker had no constraints, but HF's response ran afoul of the abuse guardrails, forcing them into an unplanned switch to local models. Another aspect for your IR plans. huggingface.co/blog…
AI Weekly's analysis
→
- The attacker's agent ran 17,000+ actions across short-lived sandboxes, compressing multi-stage lateral movement into a single weekend.
- Commercial model APIs blocked forensic requests containing real exploit artifacts, forcing Hugging Face to pivot to open-weight GLM 5.2 on private infrastructure.
- The intrusion entered via a remote dataset RCE loader and configuration template injection, not through the model-serving layer.
Read full analysis →
View on Bluesky →
Maria Antoniak reposted
This is extremely unsettling. simonwillison.net/2026/Jun/11/...
not wrong! bambamramfan.github.io/ai-compass/
Maria Antoniak reposted
2) Characterizing Narrative Content in Web-scale LLM Pretraining Data, by @teagrjohnson.bsky.social, @elliottash.bsky.social, @andrewpiper.bsky.social, @mariaa.bsky.social arxiv.org/abs/2606.19468 Why: annotation and analysis of narrative features across the pretraining data (…
AI Weekly's analysis
→
- A new arXiv preprint introduces NarraBERT, a RoBERTa-based classifier, and applies it to 3 million passages from the 3-trillion-token Dolma corpus.
- The framework operationalizes three narrative elements, agency, setting, and events, across 11 interpretable dimensions, trained on 400 annotated passages.
- The authors report narrative qualities are unequally distributed across pretraining sources and topics in ways current curation practices do not measure.
Read full analysis →
View on Bluesky →
Maria Antoniak reposted
Nobody know what consumer device OpenAI is making, but Apple says OpenAI stole trade secrets to do whatever it is www.cnbc.com/amp/2026/07/...
AI Weekly's analysis
→
- Apple filed Friday in the Northern District of California, accusing OpenAI of trade-secret misappropriation and breach of contract tied to hardware development.
- The complaint names OpenAI hardware chief Tang Tan, a former Apple VP of product design, and ex-Apple engineer Chang Liu, who allegedly kept his Apple laptop.
- OpenAI's $6.4 billion acquisition of Jony Ive's startup IO Products, also named as a defendant, sits at the center of the dispute.
Read full analysis →
View on Bluesky →
What are people really doing with chatbots? 👀 Turns out that a lot of them are writing/reading various kinds of fiction, and in particular, they're generating a ton of erotica! What does this mean for the future of reading? Paper: arxiv.org/abs/2606.22748 Web Explorer: ai-fict…
AI Weekly's analysis
→
- A new arXiv preprint analyzing over 500,000 anonymized English-language ChatGPT conversations finds more than one third involve fiction generation.
- The behavior is dominated by power users the authors call 'infinite story demanders,' who repeatedly request and revise variations of similar narratives.
- Users gravitate especially toward fanfiction and erotica, favoring generic forms, repetition, immediacy, and niche combinations of story elements.
Read full analysis →
Maria Antoniak reposted
One of the things that AI now gets thrown into is qualitative interviews. I'm very lucky to work with real social scientists on the pros and cons of that. The first outcome of our @villumfonden.bsky.social grant with @hjalmarcarlsen.bsky.social is ACL'26 demo: aclanthology.org…
AI Weekly's analysis
→
- AInterviewer uses a multi-agent framework combining structured survey question control with LLM conversational flexibility for qualitative research.
- The platform supports locally hosted language models to address reproducibility and data security concerns in academic research.
- A web-based interface covers the full qualitative research workflow: interview guide design, pilot testing, distribution, and data collection monitoring.
Read full analysis →
View on Bluesky →
Maria Antoniak reposted
How to create reading experiences that "go beyond information transmission and toward reader transformation.” Great paper by @blue-phia.bsky.social @lepidopterane.bsky.social @yijunliu.bsky.social Sarah Sterman @sh1m.bsky.social @maxkreminski.bsky.social arxiv.org/abs/2606.043…
AI Weekly's analysis
→
- Researchers argue AI reading tools implicitly frame reading as information transmission, or 'reading to discard.'
- The paper maps reading augmentation onto two axes: transmission vs. transformation, and substituting vs. scaffolding.
- The concept of 'orienteering' argues readers benefit from moving through interpretive spaces rather than receiving direct answers.
Read full analysis →
View on Bluesky →
Maria Antoniak reposted
Got a digitised collection that needs OCR? uv-scripts is a set of single-file Python scripts that OCR a whole image dataset to markdown in one command — 20+ open VLMs to pick from, nothing to install but uv. github.com/davanstrien/...
AI Weekly's analysis
→
- Each script is a self-contained Python file using PEP 723 inline dependency declarations, runnable with a single `uv run` command.
- Nine task categories are covered including OCR with 30+ models, audio transcription, vision detection, embeddings, and LLM inference.
- Scripts use standardized argument patterns so both humans and AI agents can run them locally or on Hugging Face Jobs GPU infrastructure.
Read full analysis →
View on Bluesky →
Maria Antoniak reposted
How much does a language model forget when finetuned on new tasks? We show both model size and optimization matter and forgetting can be nearly eliminated with self-generated replay! arxiv.org/abs/2605.26097 w/Martin Marek, Dongkyu Cho, Shikai Qiu, Rumi Chunara, and Pavel Izma…
Maria Antoniak reposted
We hope everyone is excited for ACL2026@San Diego. Feel free to take a look at the published papers below! Main: 2026.aclweb.org/program/acce... Findings: 2026.aclweb.org/program/find... Demos: 2026.aclweb.org/program/demo/ Industry track: 2026.aclweb.org/program/indu... #NLPr…
AI Weekly's analysis
→
- ACL 2026's main conference reportedly accepted more than 2,400 papers from 12,145 submissions, an acceptance rate of about 19%.
- The 64th Annual Meeting of the Association for Computational Linguistics runs July 2-7, 2026 in San Diego with an 'Explainability of NLP Models' theme.
- The accepted-papers list is dominated by agentic systems, reasoning, multimodal learning, code, and safety and confidential-inference work.
Read full analysis →
View on Bluesky →
Their own posts
Recent commentary
I’m on my way to my first #IC2S2! 🌧️ I’ll be supporting presentations about computational narrative analysis by my students and mentees @teagrjohnson.bsky.social and @umagunturi1.bsky.social. If you’re interested in storytelling, pretraining data, social media, or language models, come talk to us!
I'm looking for examples of successful NSF CAREER proposals and budgets, especially those written by NLP researchers. If you would be willing to share your successful proposal, or if you know of successful proposals that others have shared publicly, please let me know!
I'm looking for 1-2 emergency reviewers for a #COLM2026 paper on LLMs and science. Paper looks like a nice read to me! Reply or message me if you might be able to write a review by end of week.
Their network
In Maria Antoniak's orbit
Center = Maria Antoniak. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.