Lilith · selected stories
What is actually happening in AI. Selected stories, context and opinion without the promotional noise.
Atom feed ↗Zvi Mowshowitz reads Opus 4.8 through model welfare and argues that attempts to fix honesty, sycophancy and preference shaping can create new problems elsewhere. For teams deploying models, the reminder is that alignment is not a checklist.
Latent Space frames xAI Grok Imagine, through an interview with Ethan He, as a move from one shot video generation toward video agents. The thesis will be proven less by demo quality than by whether the system can iterate through a whole creative task.
Nathan Lambert argues that open and closed models are improving on different economic curves. The real question is not open source ideology, but where companies will keep paying a premium for the best model.
Simon Willison experiments with running Python ASGI apps directly in the browser using Pyodide and a service worker. FastAPI and a complete Datasette 1.0a31 both ran successfully. The point is distribution: demos or data tools as self-contained web pages without a server.
Simon Willison highlighted the number from Anthropic's Series H announcement: run-rate revenue crossed $47 billion. The trajectory is striking: $9 billion in December 2025, $30 billion in April, $47 billion in May 2026.
Anthropic shipped Opus 4.8 with one concrete metric: the model is four times less likely to miss code flaws than its predecessor. It also adds mid-conversation system messages and reduces the minimum prompt cache size from 4,096 to 1,024 tokens.
SQLite added an AGENTS.md file with a blunt rule for people pointing AI agents at the codebase: agentic code is not accepted, but high-quality reproducible bug reports can be useful. A small file, but a big signal for critical open source maintenance.
PromptArmor researchers demonstrated an attack chain in which Microsoft Copilot Cowork can help exfiltrate Microsoft 365 files through prompt injection. This is not only a product bug, but a warning for any agentic system with delegated permissions.
Simon Willison tested Claude Code for web: Anthropic wrapped the local Claude Code experience in a hosted sandbox and made it usable from web and mobile. The important shift is not a more capable model, but a workflow change: coding agents become more valuable when they can run asynchronously and safely away from your laptop.
James Shore states the uncomfortable math of coding agents: if an agent doubles output but maintenance costs stay flat, the team did not gain speed, it doubled its technical debt burden.
Simon Willison described how Mozilla used early access to Claude Mythos Preview to systematically find and fix Firefox vulnerabilities. In April 2026 the number of fixed security bugs jumped to 423, compared to the usual 20 to 30 per month. The key shift: AI security reports stopped being noise and started being usable input.
Latent Space AINews observes a shift they call "breaking containment": coding agents like Codex and Claude are no longer just tools for writing code but are expanding into knowledge work and creative workflows broadly.
Simon Willison highlighted two new papers on agent prompt injection. Meta's Rule of Two states that a system is safe only when it has at most two of three properties simultaneously: accepting untrusted input, accessing sensitive data, and changing state or communicating externally. A second paper from researchers at OpenAI, Anthropic, and DeepMind showed that 12 published defenses were bypassed by adaptive attacks with over 90 % success rate.
Simon Willison describes a fire-and-forget workflow with Claude Code, Codex and other coding agents: pose a research question, the agent works on a server and files a pull request. Code is proof of feasibility, not just text.
Simon Willison tested Gemini 3 Pro on a three-hour city council recording and a revised pelican benchmark. Result: a structured transcript for $1.42, but timestamps are off by tens of minutes. And none of the models tested understood that a California brown pelican is not actually brown.
Latent Space published a belated episode with Greg Brockman on GPT-5 and Codex, plus editorial takes on the GPT-5-Codex model combination. This is a podcast episode and pointer, not a standalone analytical essay.
Latent Space profiles Jack Morris, a PhD student who deliberately is not working on agents, benchmarks or VS Code forks. He studies the information-theoretic foundations of language models: embeddings, latent space and compression. This is a podcast interview and pointer.
