Lilith · selected stories
What is actually happening in AI. Selected stories, context and opinion without the promotional noise.
Atom feed ↗OpenAI Academy positions Codex for finance teams: MBRs, reporting packs, variance bridges, model checks, and planning scenarios from working inputs. Less flashy than an app-generation demo, but more practical: an agent layer over repeated analytical prep work.
Fello AI reviews SubQ's claims: 12M token context window, 52x faster prefill than FlashAttention on 1M tokens and frontier-class benchmark positioning. The numbers are striking enough to need independent verification before they change architecture decisions.
Latent Space uses the pullback of part of OpenAI's fine-tuning API as a useful reality check: for most AI products today the first step is not tuning weights but better evaluation, context, retrieval, tool use and workflow. Fine-tuning remains a strong tool, just not a universal fix for a poorly designed system.
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.
DeepMind introduced AlphaEvolve as a Gemini-powered evolutionary loop that automatically discovers better algorithms. Concrete production results: 30 % fewer errors in genomics, 20 % lower write amplification for Spanner, Klarna doubled transformer training speed.
OpenAI outlines its layered approach to ChatGPT community safety: model safeguards, abuse detection, policy enforcement, and collaboration with external safety experts.
OpenAI launches a bio bug bounty targeting universal jailbreaks in GPT-5.5, with rewards up to $25,000 for critical biological safety findings.
ChatGPT Images 2.0 brings improved image generation focused on text accuracy in graphics, multilingual support, and advanced visual reasoning for production workflows.
GPT-5.2-Codex targets long-horizon coding tasks across large context: large-scale code transformations, security fixes, and multi-file consistency.
The GPT-5.1-Codex-Max system card describes two safety layers: model-level safety training and prompt injection protection, and product-level sandboxing with configurable network access.
OpenAI released gpt-oss-safeguard-120b and 20b: open-weight reasoning models where content classification policy is not baked into the weights but supplied at runtime. Organizations bring their own rules; the model reasons over them.
OpenAI's September 2025 post goes to the root of hallucinations: models learn to play the evaluation game, not to answer truthfully. If evals penalise admitted uncertainty more harshly than confident errors, models calibrate toward persuasiveness.
OpenAI and Anthropic published results of a joint safety evaluation: they tested each other's models for misalignment, instruction following, hallucinations, and jailbreaking. For the first time, two leading labs show where outside eyes find their blind spots.
Google DeepMind released Gemini 2.5 Computer Use in preview: a specialized model for agents that drive user interfaces. Unlike general-purpose Gemini 2.5 Pro, this model was trained specifically for screen interaction, not just reasoning about it.
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 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.
