Topic
Everything filed under AI, newest first.
RSS · JSON · All topics
Anthropic Isn't Just a Company, It's a Payoff
In BIG, guest writer Lee Hepner argues that Anthropic's planned trillion-dollar IPO, revealed in its draft S-1, would chiefly enrich Google and Amazon, and that the arrangement makes sense only as a vehicle for big tech monopolies rather than a true competitor.
11 min · 2,457 words
Appraisal by AI: Model-driven victim prioritization in the VOLTA infostealer
Cisco Talos analyses VOLTA, a self-deleting infostealer that, after stealing credentials, sessions and crypto wallets, sends a structured evidence bundle to an LLM to rank the victim's assets by monetization value, failing over across a three-model chain when a model refuses.
7 min · 1,556 words
A year after his last stock-take, Salesforce AI Research engineer Ben Stolovitz describes how his AI use has changed: his personal AI policy, heavy Claude Code use for coding, research and search, summarization, transcription and voice, writing, and art, plus the subscriptions he keeps.
21 min · 4,863 words
Anthropic's IPO: Can unprecedented revenue growth outrun 'irresponsibly large' investment?
Leon Lin reads Anthropic's S-1 filing, walking through its stated assumptions, business model, revenue, costs and cash flow, to judge whether the company's record revenue growth can outpace its enormous compute investment ahead of one of the largest AI IPOs in history.
15 min · 3,482 words
Prospero: a new coding harness for a more civilized age
Chris Wellons introduces Prospero, a minimalist coding agent harness he built for running local open-weight LLMs, explaining his design choices and how it compares with heavier, cloud-oriented coding harnesses.
8 min · 1,906 words
Vitalik Buterin argues that alongside physical minimalism we now need digital minimalism: being intentional about how we use technology, and describes his own experiments in cutting social media, leaning on local and offline tools, command-line workflows and AI-crafted interfaces instead of the default web.
13 min · 2,960 words
Postgres 19 is late, and it's AI's fault
PostgreSQL committer Robert Haas explains why PostgreSQL 19 will likely slip to November or later: LLM-driven code review and bug hunting surfaced an unusual volume of defects in new features, forcing reverts and fixes, and he argues every developer will now need to raise their game.
4 min · 897 words
A guest essay by Northwestern computer scientist Jessica Hullman for Cosmos Institute arguing that AI tools that verify answers are not enough for science: drawing on Bill Thurston's view of proof, she contends research succeeds when it advances human understanding, so AI for science should support truth-seeking, not just correctness.
14 min · 3,160 words
The one thing every college freshman should know: what 87 AI models think
We asked 87 AI models (GPT, Claude, Gemini, DeepSeek…) the same question. 33% picked Office Hours. See every answer and who dissented.
12 min · 2,810 words
The Fruits of AI Drug Abundance: Indigestion In The Near Term
Venture capitalist Bruce Booth predicts that over the next three to five years AI will create a wave of drug candidates crowded around the same well-validated targets, worsening biopharma R&D productivity before improving it, echoing the 2001 'Fruits of Genomics' warning, and argues the real prize is AI-found novel drugs for undruggable targets.
9 min · 2,135 words
Observations from Running stealmybitcoin.com
Prompted by press coverage of the AI-found PHP2Shell vulnerability, security researcher James Pavur ran stealmybitcoin.com, a public challenge server vulnerable to PHP2Shell, to test whether the bug was exploitable outside a lab and whether inexpert attackers using AI could exploit it. Out of tens of thousands of attempts, only one person won the challenge.
12 min · 2,650 words
We’ve received our first response from the Project Suncatcher prototype satellite
Google reports its Project Suncatcher prototype satellite reached a "Hello World" milestone: the first AI response from a data center class TPU in orbit, a Gemma 4 haiku downlinked over Svalbard, ahead of thermal and radiation tests and twin interconnected satellites planned for 2027.
1 min · 231 words
Rate the average student's study habits out of 10: what 86 AI models think
We asked 86 AI models (GPT, Claude, Gemini, DeepSeek…) the same question. 45% picked 5. See every answer, who searched the web, and who dissented.
10 min · 2,187 words
The Economics of Intelligence: How and when does AGI lead to economic transformation?
Google DeepMind's Alex Imas, Andrew Koh and Shane Legg argue that gains in individual AI competence shift wages and the human/AI mix gradually, while crossing a threshold in agent coordination abruptly reshapes firm scale, market concentration, employment and wages. They map four capability regimes (minimal AGI, supercompetent AI, supercooperative AI, ASI), drawing on the paper Organizing Intelligence.
17 min · 3,856 words
Mistral Large 4 (Le Chonk) is now open weight
Mistral AI releases the weights of an improved Mistral Large 4 (Le Chonk), its 1-trillion-parameter natively multimodal model, after further reinforcement learning since the public preview. The post covers gains in agentic coding and defensive cybersecurity benchmarks, three reasoning modes, enterprise and sovereignty use cases, partner quotes, the Mistral Ecosystem Open-Weight license, and new API pricing.
15 min · 3,342 words
Managed Deep Agents v0.10: Browser Use and Async Subagents
LangChain’s Managed Deep Agents v0.10 adds browser use and async subagents so orchestrators stay responsive while long tasks run in the background.
6 min · 1,267 words
When AI Writes the Code, Who Proves It Works?
Ben Lorica interviews Lean creator Leo de Moura (AWS, Lean FRO) on how AI is making formal verification more accessible, why AI-generated code and autonomous agents need machine-checked specifications and proofs, verifying Rust, autoformalization, Mathlib, and what this means for engineers and mathematicians.
33 min · 7,503 words
Why We're Moving n8n's Automation Team Beyond the Service Desk Model
Romina da Costa, head of internal AI and automation at n8n, explains why a centralized automation team that builds workflows on request eventually becomes a bottleneck with 12,000+ internal workflows to maintain, and shares the four principles her team now follows to shift from a service desk to an enablement function that helps every team build and own its automations responsibly.
10 min · 2,226 words
Claude Code burns ~15–20% of your tokens on a 3-word status line. Here's how to turn it off
Flamerix AI reports that Claude Code's agent_summary feature sends each background agent's full transcript to its model about every 30 seconds just to produce a 3-5 word status line, measuring it at about 16% of Claude Code spend, and publishes an MIT-licensed Bun preload that turns it off without modifying the binary, with install, verification, tested versions and a timeline.
7 min · 1,722 words
How we’re training a Figma-fluent editing model
Figma's Josh Clemm describes the research behind the Figma agent's first-party editing model: supervised fine-tuning on synthetic edit examples, a move toward reinforcement learning on newer base models such as Qwen, multi-dimensional design-quality evals, and how the model fits alongside third-party models in the agent.
5 min · 1,108 words