www.anthropic.com/news/fable-m...
Who's Who of AI
Stephen Turner
Associate Professor and Research Dean at UVA School of Data Science, #Rstats enthusiast, dad, runner, guitar noise-maker. Views my own.
Web: https://datascience.virginia.edu/people/stephen-turner
Newsletter: https://blog.stephenturner.us
What they're sharing
Statement on the US government directive to suspend access to Fable 5 and Mythos 5 anthropic.com
Is AI ruining our skills? Early results are in — and they’re not good nature.com
AI-Generated Papers in the NeurIPS 2026 Position Paper Track – NeurIPS Blog blog.neurips.cc
Opinion | Dear A.I. Companies: The Doom Trolling Needs to Stop nytimes.com
Claude Science, an AI workbench for scientists anthropic.com
Is AI Profitable Yet? isaiprofitable.com
Scientists have a bad case of AI FOMO, Nature poll reveals nature.com
pnas.org
Generative artificial intelligence creates delicious, sustainable, and nutritious burgers | npj Science of Food nature.com
Introducing Claude Opus 4.8 \ Anthropic anthropic.com
Hallucinated citations highest in social sciences preprints site nature.com
Disparate privacy risks from medical AI | Nature nature.com
Articles & links
www.nature.com/articles/d41...
Interesting data and analysis behind NeurIPS's new AI policy blog.neurips.cc/2026/06/02/a...
AI Weekly's analysis
→
- NeurIPS desk-rejected 178 papers (18.4%) via Pangram; 123 more must prove human authorship by June 15, 2026.
- Default Pangram window sizes flagged 42.7% of submissions; refined 100-word windows reduced that rate to 12.7%.
- Comparison venues FAccT 2025 and NeurIPS E&D 2026 showed 0% and 2.1% maximum AI scores respectively.
Read full analysis →
Good essay from Cal Newport (Deep Work, Slow Productivity, ...) www.nytimes.com/2026/06/17/o... (gift link 🎁)
Blog: www.anthropic.com/news/claude-...
isaiprofitable.com
AI Weekly's analysis
→
- The AI industry has spent roughly $1.4 trillion against $613 billion in revenue, leaving almost every major player still in the red.
- Amazon leads all spenders at $313 billion invested; NVIDIA is the sole major AI player currently in the black, profiting from hardware.
- The site's anonymous creator predicts clarity on AI viability by 2030 and notes Anthropic recently approached its first profitable quarter.
Read full analysis →
Ehhh, I wouldn't call this FOMO. "48% said that they feel negative towards AI...Only 23% of the respondents felt that AI tools were having a positive impact on research, whereas 31% said that the technology was negatively affecting science." www.nature.com/articles/d41...
The diffusion of large language models in published academic articles www.pnas.org/doi/10.1073/...
Gen AI creates delicious, sustainable, and nutritious burgers www.nature.com/articles/s41...
AI Weekly's analysis
→
Read full analysis →
Claude Opus 4.8 - if this honesty thing is real that'd be nice www.anthropic.com/news/claude-...
AI Weekly's analysis
→
- Opus 4.8 matches Opus 4.7 pricing at $5/$25/M tokens; Effort Modes replace pricing tiers as the cost-quality dial.
- Dynamic Workflows impose hard ceilings: 1,000 total subagents, 16 concurrent; workflow plans live in JavaScript variables outside Claude's context window.
- SWE-bench Pro score jumps from 64.3% (Opus 4.7) to 69.2% (Opus 4.8); the model flags its own code flaws 4x more often than its predecessor.
Read full analysis →
140,000 fake citations in 2025 alone 😱 www.nature.com/articles/d41...
Disparate privacy risks from medical AI www.nature.com/articles/s41...
AI Weekly's analysis
→
- Group-level privacy metrics in medical AI can look safe while individual rare-disease and minority patients face near-perfect re-identification risk.
- Researchers trained roughly 200 AI model versions across seven clinical datasets, finding the most vulnerable patients identifiable almost perfectly by membership inference attacks.
- As AI model capacity and expressive power grow, the number of highly exposed individual patients rises considerably alongside diagnostic performance.
Read full analysis →
Their own posts
Recent commentary
Someone please define these terms for me. 1. AI literacy. 2. AI fluency. 3. AI governance. And don’t ask AI.
IANAL, but pretty sure if you publish a preprint under CC-BY, you're effectively telling the world, any AI company, qed, whoever: "here's my paper, you have my permission to do whatever you want to with it." Which includes AI training to create a derivative product and sell it back to you.
X/LinkedIn these days: Let's dive in and delve into the rich tapestry of your load-bearing claim—rather than merely accepting it—because fundamentally you're doing zero moves there, and the gap is what's actually interesting. It's worth noting: this is a paradigm-shifting, AI-driven opportunity 1/ 🧵
I'd like to see the Claude/ChatGPT histories of all these folks booing commencement speakers off stage 🤔
Their network
In Stephen Turner's orbit
Center = Stephen Turner. 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.