The Archive

User observes Claude adding human time estimation language to task outputs; questions rationale and potential misalignment.

Figure AI's humanoid robot completes 30+ hour continuous operation task, raising labor displacement concerns.

OpenAI faces class-action lawsuit alleging unauthorized sharing of ChatGPT user data with Google and Meta.

Researcher implements automated red-teaming loop using RL to generate diverse jailbreaks against Qwen3.5, then uses discovered attacks to improve model defenses.

Llama-Studio: open-source WebUI for managing multiple local llama-server instances and configurations.

Independent researcher trained small model via self-generated synthetic data and verifiable rewards, achieving 80% HumanEval and outperforming GPT-3.5 on math.

Coding agents reduce switching costs between programming languages and frameworks, enabling rapid rewrites (React Native, Bun); language/tool lock-in is eroding.

Here's what the biggest tech court case of the year is all about.

Mitchell Hashimoto notes Bun's language-agnostic portability (Zig to Rust in 1-2 weeks) signals programming languages are now commoditized, not strategic lock-in.

Today was closing arguments in the Musk v. Altman trial, and I almost feel bad writing about the unbelievable demolition derby I just witnessed. Steven Molo, Musk's lawyer, stumbled over his words. He at one point called Greg Brockman - a co-defendant - Greg Altman. He erroneously claimed that Musk wasn't asking for money and had to be corrected by the judge. He made it clear we've heard from many liars over the past few weeks, but offered little evidence for Musk's actual legal claims. OpenAI's lawyer, Sarah Eddy, countered this by simply arranging the mountain of evidence that the company i...

Abridge reports 100M doctor visit transcriptions with 10–20 hour monthly savings and sub-minute prior authorization processing using conversational AI.

datasette-agent 0.1a1 filters table listing based on execute-sql permission for users.

More than 50 employees have reportedly left Elon Musk’s newly merged SpaceXAI since February, raising questions about burnout, leadership changes, talent poaching, and whether liquidity events weakened retention incentives.

User discusses compliance with copyright restrictions when generating images using AI models.

Empirical study finds FP8 KV-cache quantization outperforms TurboQuant variants, delivering 2x capacity with negligible accuracy loss in local inference.

Yesterday, in Musk v. Altman, before the jurors came in, Sam Altman's team passed up what looked - from a distance - like a little league trophy. It was not. Yvonne Gonzalez Rogers had the lawyers read the inscription aloud for the press: "Never stop being a jackass." It's a commemoration OpenAI employees bought for research scientist Josh Ackiam, who testified yesterday. How exactly did this come up in a trial about nonprofit contract law? Allegedly, when Elon Musk was leaving OpenAI, he talked about wanting to race ahead of Google. Achiam, who worked on AI safety, asked if that was really s...

The update gives users enhanced flexibility over how they can manage their workflows.

Reddit post about running code generation model on mobile device; insufficient detail to assess technical merit or novelty.

I was chatting with Opus 4.7 about choosing an optimal step-down IC when it suddenly tried to inject a fake system prompt into the conversation. Another time, without any prompting, it leaked what looked like part of a system prompt. This is happening more and more for me. Anyone else seeing similar behavior?

Sea Limited deploys OpenAI's Codex across engineering teams to accelerate AI-native software development in Southeast Asia.

OpenAI expands Codex API access enabling code generation from any environment or interface.

OpenAI is going to let users access Codex, its desktop AI tool that can write code and use apps on your computer, from the ChatGPT app on your phone. Following the surge in popularity for Anthropic's Claude Code, OpenAI has been working quickly to try and catch up, including by cutting back on "side quests," shutting down projects like the Sora video-generation tool, and focusing on growing its enterprise business. The company's push included the recently released major update for Codex that lets it operate apps on macOS - a potentially major step as part of its ambitions to make a desktop "s...

NVIDIA reportedly plans price increases for RTX 5090 and RTX 50 series GPUs due to rising GDDR7 memory costs.

Richard Socher's new $650 million startup wants to build an AI that can research and improve itself indefinitely — and he insists it will actually ship products.

Anthropic partners with Gates Foundation on $200M initiative; funding and strategic alignment unclear.

Mythos Preview LLM helped develop macOS kernel exploit for Apple M5 in 5 days, demonstrating AI-assisted security vulnerability discovery.

And it doesn't show until two steps in. Went through 80% of my usage overnight. I don't know if it was happening before 2.1.139, but that's what I was running. FFS, Dario. Just put me on the payroll. And in the meantime, SORT YOURSELVES OUT.

Agentic inference has fundamentally changed the runtime dynamics of inference workloads by introducing non-deterministic trajectories—actions, observations,... Agentic inference has fundamentally changed the runtime dynamics of inference workloads by introducing non-deterministic trajectories—actions, observations, and decisions that an AI agent produces while working through a task. These trajectories compound end-to-end latency across hundreds of inference requests per session. NVIDIA Vera Rubin NVL72 handles the bulk of that inference load as… Source

Town’s 49,000 California residents compete with Nevada data centers for energy.

According to Bloomberg, OpenAI has enlisted an outside law firm to work through its options.