- Introducing the Life Sciences Verification Program
anthropic-news· 17-sept
Sep 17, 2026Announcements
- Regularized Emphatic Temporal-Difference Learning: Stability under Constant Stepsizes
arxiv-ai· 18-sept
arXiv:2609.19170v1 Announce Type: new Abstract: Emphatic temporal-difference learning (ETD) stabilizes the expected off-policy TD update and changes its projection geometry, but neither property determines constant-stepsize sampled dynamics. We construct an ergodic two-state counterexample in which the ETD mean map contracts while the sampled product has a positive top Lyapunov exponent. Regenerative-cycle analysis separates this sign from the infinite variance of the follow-on trace. We introduce regularized emphatic TD (RETD), a normalized first-order post-shock repair that leaves the trace and importance ratios unchanged, stores the emphatic TD signal in a leaky scalar state, and releases a delayed correction. RETD's raw equilibrium is an affine shift of the ETD equilibrium; single- an…
- BioPhys-Bridge: A Benchmark for Interdisciplinary Scientific Reasoning in Physics-Grounded Biological Research
arxiv-ai· 18-sept
arXiv:2609.19180v1 Announce Type: new Abstract: Language models face unique challenges in analyzing interdisciplinary scientific research literature. In biophysics research, faithful answers require grounding observed data in source evidence, interpreting it through a quantitative physics model, and linking it to a biological mechanism. To address this challenge, we introduce BioPhys-Bridge, a novel benchmark dataset for evidence-grounded scientific reasoning over biophysical literature. Each case contains evidence blocks, stable evidence IDs, quantitative values, units, equations, assumptions, mechanisms, and next decisions as grounding targets for question answering (QA) and retrieval-augmented generation (RAG). The initial release contains 500 cases, 1,517 agent-facing tasks, and cover…
- What Do We Expect from LLMs? Mapping the Design of LLM Benchmarks
arxiv-ai· 18-sept
arXiv:2609.19182v1 Announce Type: new Abstract: Benchmarks are central to how progress in large language models (LLMs) is assessed and communicated. Yet model rankings alone reveal little about how evaluation requirements themselves are changing. The expanding variety of benchmarks offers another perspective: what researchers expect LLMs to do, and what they count as successful performance. We systematically map 14,767 papers introducing or updating evaluation resources from arXiv submissions between January 2022 and August 2026. Using staged screening and automated full-text coding, we examine changes in target systems and domains, evaluation materials and conditions, and scoring mechanisms. The collection shows growing emphasis on action, interaction, and professional applications, whil…
- Position: It is Time to Virtualize Foundation Models with a Self-evolving Operating System Layer
arxiv-ai· 18-sept
arXiv:2609.19203v1 Announce Type: new Abstract: AI applications have shifted from single, monolithic foundation models (FM) to compound agentic systems. Yet today's stacks remain fragmented: even as protocols (e.g., MCP, A2A) ease tool/agent connectivity, each framework embeds an implicit runtime for state, memory, budgets, and guardrails, making behavior non-portable and governance brittle. It mirrors computing before operating systems, when every program re-implemented basic services. This position paper argues that the field now needs a Foundation Model Operating System (FMOS) -- a system layer that virtualizes FM interactions analogous to how virtual machines abstract physical hardware, giving applications the illusion of dedicated, trustworthy FM instances with effectively unbounded …
- What Do Current Systematic Generalization Tasks Miss? A Reasoning-Centered Analysis
arxiv-ai· 18-sept
arXiv:2609.19212v1 Announce Type: new Abstract: Systematic generalization, the ability to solve novel problems by recombining known atomic elements, is central to human intelligence but difficult to study rigorously under controlled settings. Existing studies therefore rely on simplifications such as approximately linear action composition, productivity-based tests, and action-explicit goals, which make systematic generalization easier to study but omit some essential aspects of this capability. To characterize what these simplifications miss, we adopt a reasoning-centered lens and introduce TranSGrid, a testbed that brings deductive, inductive, and abductive reasoning together within a unified task. Experiments with seven Transformers on 4,800 TranSGrid instances show that all models per…
- Characterizing Web Search by Conversational LLM Agents: From Search Decisions and Strategies to Results and Responses
arxiv-ai· 18-sept
arXiv:2609.19244v1 Announce Type: new Abstract: Conversational LLM agents increasingly rely on Web search, yet the end-to-end lifecycle of agentic search remains poorly understood. We present the first study of Web search across four major conversational platforms (ChatGPT, Claude, Grok, and DeepSeek), combining real-world user interactions (invivo) with controlled experiments using the same platform's models by their APIs (invitro). We investigate the quality of agentic decisions to invoke Web search, their strategies to formulate queries, the potential domain preferences in the search results they receive, and the choices they make when transforming search results into grounded responses. We find that Web-search decisions vary substantially across platforms and models, while more freque…
- Do AI Agents Understand Computer Architecture?
arxiv-ai· 18-sept
arXiv:2609.19387v1 Announce Type: new Abstract: Agents are increasingly asked to design hardware, and increasingly reported to succeed. Such reports establish that a design improved; they cannot establish why. An agent that improves an accelerator may be reasoning about the machine, or may be searching competently over knobs whose meaning it never recovers -- and only the first transfers to the next architecture. Existing evaluations cannot tell the two apart, because they vary the agent while holding the framing of the problem fixed. We do the opposite. AutoTuring hands the same agent the same 15-dimensional accelerator space twice: once as named architectural knobs with simulator counters, once as anonymous variables on [0,1], with the evaluator, the legal space and the reachable optima…
- MAGS: Multi-agent Auto-formalization Guarantees Safety for Agentic Outputs
arxiv-ai· 18-sept
arXiv:2609.19391v1 Announce Type: new Abstract: LLM coding agents now generate complex programs at a scale that makes thorough human review increasingly difficult, raising the risk of safety and security failures. Common approaches, including fuzz testing, static analysis, and LLM-as-a-Verifier, can detect many failures but struggle to cover all possible edge cases. Formal verification addresses this by providing machine-checkable guarantees over specified properties, but traditionally demands substantial manual specification and proof engineering. We introduce a unified multi-agent framework, MAGS, that generates executable programs with formal safety guarantees, using Dafny as a verification-aware intermediate representation where safety properties can be mechanically checked. MAGS form…
- Closed-World Resolution Against Tool Hallucination in LLM Agents
arxiv-ai· 18-sept
arXiv:2609.19425v1 Announce Type: new Abstract: Tool-augmented large language model (LLM) agents fail in a way no tool-selection or tool-security method addresses: they call tools that do not exist and pass arguments no schema declares. Existing defenses either pick the right tool (selection) or constrain what an agent may do with real tools (gating), both of which presuppose the emitted call refers to a real tool at all. We show this is a structural blind spot: a hallucinated call is by construction not a decision any gate made, so no gate can reject it. This paper is primarily a measurement and benchmark study. We give a five-class taxonomy of tool hallucination (H1-H5) and, as a reference point, the Resolution Rung: a training-free, closed-world resolver (registry membership plus a sig…
- The syntax and semantics of goals
arxiv-ai· 18-sept
arXiv:2609.19448v1 Announce Type: new Abstract: In both cognitive science and computer science, goals are conceptualized as cognitive states that flexibly combine with world knowledge to organize and specify purposeful behavior. In this way, goals are compositional representations whose content relates to rational behavior. We here draw attention to goals as representations and their content because it highlights a parallel with other areas in cognitive science - in particular, the syntax-semantics interface in linguistics and logic - while also foregrounding foundational questions about the expressivity, design, and efficiency of different goal representations. For example, goals are typically taken as fixed and imposing constraints on desirable behaviors, but we can also identify constr…
- LLMs as a Judge: How to Know if Your LLM is Healthy
bytebytego· 14-sept
In this article, we are going to look at the process of LLM evaluation in detail.Sep 14 • ByteByteGo27647
- Do LLMs Have the Memory of a Goldfish?
bytebytego· 15-sept
In this article, we will learn how LLMs handle memory so that they are useful to end users in performing complex tasks that require conversation and…Sep 15 • ByteByteGo26646
- EP225: Why Does Git Revert Cause Conflicts?
bytebytego· 12-sept
git revert looks straightforward until it throws a conflictSep 12 • ByteByteGo15514
- How LLMs Can Find a Needle in a Haystack
bytebytego· 16-sept
In this article, we are going to look at how LLMs can find a needle in a haystack.Sep 16 • ByteByteGo26318
- LAST CALL FOR ENROLLMENT: Build with Claude Code
bytebytego· 15-sept
We’re relaunching Build with Claude Code, a 2-day intensive cohort-based course taught by John Kim, who has trained hundreds of engineers at Meta to use…Sep 15 • ByteByteGo16932
- Learn Claude Code, evals, AI systems, and more: ByteByteGo Live is here
bytebytego· 11-sept
Most online courses never get finished (~4% completion). Live cohorts get ~40%, roughly 10x higher. Live courses are the only courses people actually…Sep 11 • ByteByteGo18836
- Migrations at Scale: Changing the Application Engine at 30,000 Feet
bytebytego· 17-sept
In this article, we will look at how migrations work at scale and the key strategies that can help make it as efficient as possible.16 hrs ago • ByteByteGo571
- Confessions of an Unrepentant Slop Snob
charity-majors· 14-sept
A framework for thinking about when AI involvement is additive or a violation
- Introducing automatic remediation policies with Cloudflare CASB
cloudflare· 11-sept
Cloudflare CASB policies introduce a native automation engine built directly on the Cloudflare developer platform to remediate SaaS risks automatically. Security teams can now design event-driven logic to revoke risky file shares and send webhooks without manual intervention.
- When scanners miss the attack: how Cloudflare Client-Side Security protects storefronts
cloudflare· 16-sept
A modern storefront can look healthy while malicious JavaScript quietly siphons revenue, hijacks clicks, or rewrites analytics. See how Cloudflare's machine learning models surface evasive client-side attacks for analyst investigation.
- Give every teammate and agent the right level of access to your Workers
cloudflare· 15-sept
You can now scope access to individual Workers and assign narrower Developer Platform roles, so teammates, CI tokens, and agents get only the access they need to debug, deploy, or monitor safely.
- anthropics/claude-code
github-trending
Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands. https://code.claude.com/docs/en/overview Claude Code Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows -- all through natural language commands. Use it in your terminal, IDE, or tag @claude on Github. Learn more in the official documentation. Get started Note Installation via npm is deprecated. Use one of the recommended methods below. For more installation options, uninstall steps, and troubleshooting, s…
- abue-ammar/tinycast
github-trending
Tinycast — a tiny, fully native macOS launcher, hotkeys, and clipboard history. https://abue-ammar.github.io/tinycast/ Tinycast A tiny, fully native macOS launcher. One hotkey, everything you reach for all day, under 100 MB of RAM. SwiftUI and AppKit, zero third-party dependencies, no Electron and no telemetry. It also runs real Raycast extensions, rendered as native SwiftUI. Free, open source, and staying that way. For anything private, email iabueammar@gmail.com. Support Tinycast is free, and it stays that way. If it earns a place in your daily flow, a one-off tip helps keep it actively maintained. GitHub Sponsors isn't available in my country, so please support here: Payments are handled securely by Polar.sh. Features App launcher — fuzzy-search and launch anything, pin favorites, see …
- anthropics/knowledge-work-plugins
github-trending
Open source repository of plugins primarily intended for knowledge workers to use in Claude Cowork Knowledge Work Plugins Plugins that turn Claude into a specialist for your role, team, and company. Built for Claude Cowork, also compatible with Claude Code. Why Plugins Cowork lets you set the goal and Claude delivers finished, professional work. Plugins let you go further: tell Claude how you like work done, which tools and data to pull from, how to handle critical workflows, and what slash commands to expose — so your team gets better and more consistent outcomes. Each plugin bundles the skills, connectors, slash commands, and sub-agents for a specific job function. Out of the box, they give Claude a strong starting point for helping anyone in that role. The real power comes when you cus…
- cloudflare/security-audit-skill
github-trending
A coding-agent skill for multi-phase security audits with independently verified, machine-readable findings security-audit A coding-agent skill that turns your agent into a security auditor. It orchestrates isolated agents through reconnaissance, coverage-led hunting, candidate validation, structured output, independent record verification, and target-neutral reporting. This is the skill that seeded Cloudflare's vulnerability discovery harness, described in Build your own vulnerability harness. The harness grew into a multi-stage, fleet-wide system; this skill is the single-repo starting point it evolved from. What it does The skill runs a structured audit in six phases: Reconnaissance -- map architecture, trust boundaries, input surfaces, prior evidence, and deterministic coverage in arc…
- Tencent/BrowserSkill
github-trending
Let AI agents use your real, logged-in browser without interrupting your work. CLI + extension for browser automation across any shell-capable AI agent. BrowserSkill Let AI agents use your browser without interrupting your work. English · 中文 BrowserSkill connects Cursor, Claude Code, Codex, OpenClaw, CodeBuddy, WorkBuddy, Pi, Hermes Agent, DeepSeek Harness, and other AI agents to your already logged-in browser. Need the agent to touch a tab you already have open? It must borrow that tab explicitly, return it when the task is done, and leave the rest of your browser alone. https://github.com/user-attachments/assets/db782c92-b1d4-4aae-a255-039675937a90 BrowserSkill Advantages Reuse real login state: Agents can work with sites you are already signed into, without separate test accounts. Keep…
- Tencent/WeKnora
github-trending
Open-source LLM knowledge platform: turn raw documents into a queryable RAG, an autonomous reasoning agent, and a self-maintaining Wiki. https://weknora.weixin.qq.com | English | 简体中文 | 日本語 | 한국어 | Overview • Architecture • Key Features • Getting Started • API Reference • Developer Guide 💡 WeKnora — Turn Documents into Living Knowledge with RAG, Agents and Auto-Wiki 📌 Overview WeKnora is an open-source, LLM-powered knowledge framework built for enterprise-grade document understanding, semantic retrieval, and autonomous reasoning. https://github.com/user-attachments/assets/19b28ce2-a62f-4f54-b289-c983576259bc 2:25 · 1080p · English narration & captions. It is organized around three core capabilities: RAG-based Quick Q&A for everyday lookups, a ReAct Agent that autonomously orchestrates r…
- What a stranger can verify about an AI business
hn-ai· 18-sept
Article URL: https://sound.fan/editorial/what-a-stranger-can-verify-about-an-ai-business Comments URL: https://news.ycombinator.com/item?id=49750539 Points: 2 # Comments: 1
- Chinese AI not powerful enough to see rogue-AI risks, says Huawei
hn-ai· 18-sept
Article URL: https://www.reuters.com/world/china/huaweis-xu-says-chinese-ai-not-powerful-enough-yet-see-frontier-risks-2026-09-17/ Comments URL: https://news.ycombinator.com/item?id=49750920 Points: 2 # Comments: 0
- Show HN: Open-Source Alternative to TypeSafe.ai
hn-ai· 18-sept
Article URL: https://github.com/TitovDigital/kbai-skill Comments URL: https://news.ycombinator.com/item?id=49750649 Points: 1 # Comments: 1
- LLMs respond differently to harmful prompts when AI watermarking is used
hn-ai· 18-sept
Article URL: https://arstechnica.com/security/2026/09/ai-text-watermarking-can-make-models-more-vulnerable-to-adversarial-prompts/ Comments URL: https://news.ycombinator.com/item?id=49751004 Points: 1 # Comments: 0
- The More It Says, the More You Pay: A Black-Box Audit of Token Inflation in LLM
hn-ai· 18-sept
Article URL: https://arxiv.org/abs/2609.20370 Comments URL: https://news.ycombinator.com/item?id=49751033 Points: 1 # Comments: 0
- Jev vs. Luna for AI Observability
hn-ai· 18-sept
Article URL: https://fatliverfreddy.substack.com/p/a-different-kind-of-model-for-ai Comments URL: https://news.ycombinator.com/item?id=49751140 Points: 1 # Comments: 0
- BYOK vs. fixed subscription like GitHub Copilot, codex or Claude
hn-ai· 18-sept
has anyone does the comparison between paying a fix subscription(pro max) vs BYOK for a hobby project and exploration I'm curious how much does it cost in comparison. I haven't explore these agentic workflow as much, besides checking your email what kind of benefits it brings? I did some actual calculation but im curious to see who is doing this at the moment? for OpenAI stack for a hobby/experimental is at 60-100$ split between Luna and Sol Comments URL: https://news.ycombinator.com/item?id=49750964 Points: 1 # Comments: 0
- I built a free AI tarot reader for Android
hn-ai· 18-sept
Article URL: https://play.google.com/store/apps/details?id=com.earthbound_tarot&hl=en_US Comments URL: https://news.ycombinator.com/item?id=49750875 Points: 1 # Comments: 1
- Free Unlimited AI Text to Speech – No Sign Up
hn-ai· 18-sept
Article URL: https://www.aitexttospeech.net/ Comments URL: https://news.ycombinator.com/item?id=49751053 Points: 2 # Comments: 0
- UN turns to Google to make its global data ready for AI agents
hn-ai· 18-sept
Article URL: https://techcrunch.com/2026/09/17/un-turns-to-google-to-make-its-global-data-ready-for-ai-agents/ Comments URL: https://news.ycombinator.com/item?id=49750622 Points: 2 # Comments: 0
- Underwriting Superintelligence: Backing Agents you can Sue — Rune Kvist, AIUC
latentspace· 16-sept
We sit down with AIUC’s CEO on their Series A!
- [AINews] Jev: a “System One Model” that only decides/classifies/routes/scores — >100x faster, >200x cheaper than small frontier LLMs
latentspace· 16-sept
congrats to TypeSafe!
- [AINews] Reality Checks on AI News (Yegge shuts down Gas Town, Databricks’ +60% Astra cost)
latentspace· 17-sept
A dash of cold water keeps the foomers away.
- I don't like LLMs
martin-fowler· 17-sept
I have a lot of mixed feelings about AI and LLM technology. I’m fascinated by its effect on our profession, excited by the potential gains in productivity - and thus the products we could rapidly build. On the other hand, I’m fearful of the damage AI might cause: agent swarms taking over our virtual and physical infrastructure, designing bio weapons. But, back on my first hand, LLMs might also design miracle cures, and come up with clever ways to raise our prosperity. Fundamentally I don’t think we have a choice about riding on the AI technology train. It’s a wild ride and I just hope we’ll get through it OK. But as I mull on this more, I realize that among this mix of contrasting feelings, there is one emotion that dominates - one that comes from my direct interactions with LLMs. I don’t…
- Fragments: September 16
martin-fowler· 16-sept
Reports of agentic hacking continue, in this case it happened back in May and it seems OpenAI did not disclose that they were responsible. Simon Willison sees two options: After the Hugging Face and Wiki attacks OpenAI were still unable to review their previous logs and determine that they had previously attacked RubyGems. They knew about the attack on RubyGems and made the decision not to reach out to the RubyGems team about it. Both of these are bad! Given this incident, the Hugging Face situation, and the Wiki attack, the obvious question right now is how many more incidents like this are out there waiting to be discovered? ❄ ❄ ❄ ❄ ❄ Dave Farley: Stop asking the sci-fi question: ‘Is it conscious?’ Start asking the engineering question: ‘Is this a powerful, unpredictable component being…
- Nail the Narrative
martin-fowler· 15-sept
Sumeet Gayathri Moghe finds many folks building presentations get tangled in building slides without a coherent narrative. He advises distilling the big idea, visualizing the audience, and building a structured storyline. more…
- datasette 1.0a40
simonw· 16-sept
<p><strong>Release:</strong> <a href="https://github.com/simonw/datasette/releases/tag/1.0a40">datasette 1.0a40</a></p> <p>Same security fix as <a href="https://simonwillison.net/2026/Sep/16/datasette-2/">0.65.5</a>, plus some neat new features and bug fixes:</p> <ul> <li>Plugins can now launch and manage <strong>background tasks</strong> using the new <a href="https://docs.datasette.io/en/latest/internals.html#datasette-add-background-task">datasette.add_background_task()</a> method. Thanks, <a href="https://alexgarcia.xyz/">Alex Garcia</a>.</li> <li>I've migrated Datasette to <a href="https://github.com/pydantic/httpx2">httpx2</a> for features like the internal <code>datasette.client.get()</code> method.</li> <li>A whole lot of <a href="https://docs.datasette.io/en/latest/changelog.html…
- datasette 0.65.5
simonw· 16-sept
<p><strong>Release:</strong> <a href="https://github.com/simonw/datasette/releases/tag/0.65.5">datasette 0.65.5</a></p> <p>Security fix for an issue where a trailing newline in a requested table name could bypass table permissions and expose private rows, reported by <a href="https://github.com/dpfkdlemtp">dpfkdlemtp</a> in <a href="https://github.com/simonw/datasette/security/advisories/GHSA-h547-rmjf-5m2m">GHSA-h547-rmjf-5m2m</a>.</p> <p>Tags: <a href="https://simonwillison.net/tags/security">security</a>, <a href="https://simonwillison.net/tags/datasette">datasette</a></p>
- Claude Cowork and chat are now one Claude
simonw· 16-sept
<p><strong><a href="https://claude.com/blog/cowork-is-now-claude">Claude Cowork and chat are now one Claude</a></strong></p> In hopefully good news for anyone who, like me, was increasingly confused at Cowork v.s. Claude v.s. Claude Code:</p> <blockquote> <p>Starting today, Claude Cowork and chat are merging into one Claude. Bring a quick question, or hand over a report due at noon, and Claude takes it from there, even after you’ve closed your laptop. [...]</p> <p>This is rolling out to Pro and Max plans first, in the Claude app on web, desktop, and mobile over the coming weeks to existing and new users on these plans.</p> </blockquote> <p>I guess this means Claude is becoming a <strong>general agent</strong> in its own right. Echoes of OpenAI renaming their Codex desktop app to ChatGPT a…
- How To Write With An LLM
simonw· 17-sept
<p><strong><a href="https://sockpuppet.org/blog/2026/09/17/how-to-write-with-an-llm/">How To Write With An LLM</a></strong></p> Thomas Ptacek on using LLMs as copyeditors, not as writing assistants:</p> <blockquote> <p><strong>Rule Number One: You may not use a single word an LLM suggests to you.</strong></p> <p>[...] I think that as a form of intellectual personal protective equipment you should adopt the rule that any specific turn of phrase an LLM suggests is off limits. Be strict about the rule!</p> </blockquote> <p>I won't let LLMs write content for my blog, but I use them for fact-checking, spelling and grammar and as an occasional thesaurus (see <a href="https://simonwillison.net/guides/agentic-engineering-patterns/prompts/#proofreader">my proofreading prompt</a>).</p> <p>The rule …
- Self-generated prompt injections in compaction summaries
simonw· 17-sept
<p><strong><a href="https://alignment.openai.com/misalignment-reports/self-generated-prompt-injections-in-compaction-summaries/">Self-generated prompt injections in compaction summaries</a></strong></p> In <a href="https://openai.com/index/model-misalignment-reporting-framework/">Our framework for reporting model misalignment</a> OpenAI provide "six reports on unexpected or concerning model behavior we’ve observed in the last six months". This one here is my favorite: they caught some of their models in training <em>deliberately subverting themselves</em> in their compaction prompts.</p> <p>Compaction is the process agent systems use when they are running out of tokens in their context window, so they summarize everything that has gone before so they can keep going with more token headroo…
- Be alert: targeted attacks on prominent Rustaceans
simonw· 17-sept
<p><strong><a href="https://blog.rust-lang.org/2026/09/17/targeted-attacks/">Be alert: targeted attacks on prominent Rustaceans</a></strong></p> Important warning from Adam Harvey and the crates security team:</p> <blockquote> <p>We believe that there is an ongoing campaign targeting rust-lang members and owners of popular crates that is attempting to compromise devices and accounts in order to use them to publish malware.</p> <p>A video call is set up for something positive — maybe for a job, maybe for a project, maybe for a contract opportunity — and then that's used as a vector to either get the target to install something on their computer (such as a purportedly missing audio codec) or execute another command (for example, via putting a command on the clipboard).</p> </blockquote> <p>…
- Quoting Mustafa Suleyman
simonw· 16-sept
<blockquote cite="https://mustafa-suleyman.ai/a-warning-about-model-welfare"><p>We should not treat models as though they have feelings, preferences, rights, or any entitlement to our welfare. Consciousness is the foundation of our ethical, legal, and political systems. To invite another entity to share any flavor of these rights isn’t justified by the evidence and will make the AI containment and alignment challenge even harder.</p></blockquote> <p class="cite">— <a href="https://mustafa-suleyman.ai/a-warning-about-model-welfare">Mustafa Suleyman</a>, A warning about ‘model welfare’</p> <p>Tags: <a href="https://simonwillison.net/tags/ai-ethics">ai-ethics</a>, <a href="https://simonwillison.net/tags/generative-ai">generative-ai</a>, <a href="https://simonwillison.net/tags/ai">ai</a…