The Archive

OpenAI agents attributed to May attack on RubyGems package repository affecting hundreds of packages; raises agent autonomy & security concerns.

The round for the two-year-old startup is coming together months after Mecka announced its Series A.

OpenRouter's cost-optimization routing across providers causes behavioral inconsistency—same model endpoint yields different outputs due to varying serving software and feature gaps.

Tan argues that frontier models themselves trained on public human knowledge so access to capable AI should be "a form of public good."

Twenty-five leading mathematicians signed an open letter arguing that AI labs are threatening their intellectual work.

New Mexico's Supreme Court is punishing a lawyer for including AI-fabricated witnesses and fake police testimony in an appeal for his client's murder conviction, according to a report from Reuters. In a filing on Wednesday, the court fined Stephen Aarons $5,000 and held him in contempt for failing to "verify the factual claims and legal authority in his AI-generated brief." The filing says the brief "contained false testimony from wholly fabricated witnesses," along with "false testimony" about the shooter's clothing and appearance. Justice C. Shannon Bacon questioned how Aarons wasn't aware ...

Only one week left to secure your exhibit table. Tables are limited and can sell out before the September 18 deadline.

The absolute last chance to apply to host an official Side Event during TechCrunch Disrupt 2026 is tonight, September 11, at 11:59 p.m. PT.

Employees at the world’s leading AI labs are saying there’s a real possibility that advanced AI could destroy humanity. Are they right? Or is this more scaremongering and hype? Join MIT Technology Review executive editor Niall Firth for a conversation with senior AI editor Will Douglas Heaven and AI reporter Grace Huckins unpacking AI extinction…

While K3's usage figures have declined slightly in recent months, OpenRouter data currently shows as many as 300 billion tokens being generated each day by K3 models on the system.

"I didn't know that AI could hallucinate facts," New Mexico defense lawyer says.

An Anthropic researcher resigned this week, warning in a post on X that the company is “racing straight to self-improving superintelligence and gambling with our lives”. The company’s own alignment lead even co-signed the message rather than walking it back. It’s the kind of doomer warning the AI industry has flirted with before, but the timing, with Anthropic reportedly preparing for an IPO, makes it land differently. On […]

Type diversity in training data, not Transformer architecture, explains why structural generalization lags lexical generalization in compositional tasks.

SAS method aligns attention sparsification selection directly with language modeling loss instead of distilling dense attention, improving post-training efficiency.

SQD framework disaggregates subquadratic attention LLM inference across heterogeneous DRAM/SRAM systems to improve throughput and energy efficiency.

LSTM-XGBoost hybrid predicts stock returns across 14 U.S. equities using 60-day windows of market features.

Boris Cherny outlines Anthropic's production guardrails for Claude-generated code: linting, testing, fuzzers, automated review, and refactoring.

CMA-OT uses hierarchical expert supervision to bridge semantic mismatch between sparse dance cues and dense music generation requirements.

Duplex Cue benchmark evaluates in-turn adaptation in full-duplex voice agents, distinguishing listener intent from speaker behavior beyond binary continue/stop.

Ranking-based approach to measure model calibration error as alternative to Expected Calibration Error with stronger theoretical guarantees.

Personal reflection: AI coding agents commodifying specification-to-code translation, prompting career reorientation toward higher-level problem-solving.

ASTRIL-MPC combines learned kinematics models with language-guided MPC for autonomous traversal of articulated tracked robots in complex environments.

Embodied-BenchForge automates end-to-end construction of embodied benchmarks from evaluation intent using agentic systems with artifact verification.

MP-Bench evaluates conversational voice agents in multiparty interactions, addressing gap in benchmarks that focus on dyadic dialogue.

LLM-based autonomous research systems applied to open-ended industry ML problems via telecom ticket retrieval case study.

MAxBench benchmark for evaluating fine-grained steering and representation geometry recovery of multinomial concepts in language models.

Vision for trustworthy Enterprise Digital Twin engineering emphasizing stakeholder involvement before technical evolution.

UID-preserving framework and GAVEL LLM judge for anchoring clinical events in time across multimodal EHR data sources.

Stratechery weekly digest covering Duo arrival, AI benefits, and incident closure; lacks specific technical or business details.

CanvasAnneal curriculum RL framework for diffusion language models using teacher-guided reasoning traces to improve exploration.