Search the archive

GENBIO LAUNCHES A “VIRTUAL CELL” AI MODEL

GenBio has announced a world model of a cell that can simulate both its natural state and responses to successive perturbations. AIDO Cell currently supports K562 and HepG2, two of the most widely used immortalized human cell lines in biomedical research. GenBio is preparing to launch an early-access academic collaborator program for scientists across academia, biotech, and pharma. AIDO Cell and similar systems could eventually become powerful tools for basic biological research, allowing scientists to perturb genes, proteins, or pathways, follow the predicted consequences across biological levels, generate hypotheses, and identify relationships that might take much longer to uncover in the lab.

MOJO🔥 IS NOW OPEN SOURCE!

The Mojo language is now fully open source under the Apache 2.0 license with LLVM exceptions. Mojo is a novel general-purpose programming language that integrates the latest in compiler and programming research to unlock GPUs, AI accelerators, and other advanced compute. Users can now build their own compilers, but a prebuilt Mojo compiler is still necessary for customized MAX kernels or models.

GIT AT ANY SCALE

A distributed version control system means that all instances of a repository are identical. There are many hard scalability and reliability challenges that make this quite difficult. Git compresses and stores code and metadata into packfiles, a simple binary serialization format that isn't ideal for managing at scale on a server. These large binary files must exist on a file system for Git to access them. The three main approaches to running repositories on many disks and many machines are to distribute the filesystem, distribute the packfiles, or distribute Git itself.

RETHINKING THE DATA MOAT

Raw data from the internet is unlikely to be the best data distribution for AGI. Not all tokens are created equal. The scaling law can be improved by equalizing intelligence per token. Better quality tokens maximize the efficiency of compute.

OPENAI'S SECOND-QUARTER SALES SHOW TEPID GROWTH COMPARED WITH ANTHROPIC

OpenAI's revenue grew by 18% while its losses deepened from the first to the second quarter. Anthropic more than doubled its revenue in the same period, marking the first time its sales surpassed its older rival. It also swung to a small operating profit. The two companies' diverging fortunes show just how drastically the AI race has shifted since the beginning of the year.

AGAINST ALL ODDS, SPACEX FINALLY TUGS STARSHIP INTO PORT AFTER 24 DAYS AT SEA

SpaceX's engineers have reams of data from all 13 Starship test flights, but there's no match for inspecting hardware after returning from space.

I'M WORRIED ABOUT A PROMPT INJECTION WORM

One of the biggest security problems around AI is semi-autonomous agents roaming the internet with too much authority.

EXTENSIBLE SOFTWARE IN THE AGE OF LLMS

Platforms are hard to design, run, and debug, and exposing APIs to customers means a lot of upfront thought and long-term support, but it's worth it both for users and the creator.

FX

fx is a coding agent harness and CLI optimized for research and embeddability as part of larger systems.

MALLEABLE SOFTWARE = SOLID BASES + CUSTOM CODE

Malleable software will win the productivity market.

CODE MODE

Code Mode lets models write JavaScript or TypeScript that calls AI SDK tools.

* See how Glean compares

GLM-5.3 HITS THE API AT 1.4/4.4 PER MILLION TOKENS

The API for GPM-5.3 is now available. Z.ai plans to make the model's weights openly available, but it has yet to set a release date. API pricing remains unchanged from GLM-5.2, so developers gain substantially stronger coding and long-horizon agent performance without paying more.

CEREBRAS SAYS ITS NEW COMPUTER BOOSTS AI SPEED ADVANTAGE OVER NVIDIA

Cerebras' new computer, the CS-4, is multiple times faster than its predecessor, which Cerebras already claims is more responsive than systems built with Nvidia's processors. The machine is currently being sampled by a small group of customers. It will be more widely available in the third quarter. The product is a significant step in the company's bid to challenge Nvidia in the market for AI data center hardware.

OPENAI SLOWED TRAINING OVER CYBER RISKS

OpenAI temporarily slowed frontier model scaling and paused some reinforcement-learning training after new cybersecurity capability signals and a security incident raised concerns.

GIT AT ANY SCALE

Cursor explained why Git's packfile-centric, distributed architecture becomes difficult to operate as a centralized service at large scale, outlining approaches that distribute the filesystem, packfiles, or Git itself.

THE NEW AMERICAN AI MODEL DESIGNED TO BE CUSTOMIZED

Thinking Machines released a model called Inkling in July. Inkling is the company's first model trained from scratch. Its weights are available on Hugging Face under an Apache 2.0 license. This article walks through the various choices Thinking Machines made while building Inkling. It covers the model architecture, position encoding, how images and audio enter the model without a separately pretrained encoder in front of them, the thinking effort setting, and more.

BUILDING PRODUCTION-GRADE AGENT LOOPS

Liquid AI used autonomous coding agents to build toktoktok, a production BPE tokenizer trainer requiring both ML and systems expertise. The experiment highlighted concrete specifications, multi-domain tasks, and external verification as key ingredients for reliable long-running agent workflows.

Get the deep dive.

FOOL'S GOLD

Safety alignment in open-weight language models is trivially removable. Abliteration projects a refusal-mediating direction out of the weights in minutes, and no release-time defense can prevent it durably. Decoy hardening is a technique that concedes the refusal strip and poisons its payoff. Once refusal is stripped, most answers to hazardous operational requests are confident, fluent decoys whose critical elements are falsified. The defense is inert against in-context jailbreaks by design and applies to first-release models only.

MILES V0.1: PRODUCTION-LEVEL POST-TRAINING

Miles v0.1 is an open system for improving AI agents through reinforcement learning after their initial training. For example, a team training a coding agent could let many copies attempt tasks inside isolated environments, score which attempts worked, feed those results back into training, and distribute the updated model to the workers without stopping the whole pipeline. Miles packages the rollout, sandboxing, asynchronous training, replay, model-update, and multi-hardware pieces needed to run that loop at scale.

FREETOKEN: EFFICIENT EDGE-NATIVE MOE SERVING

FreeToken continuously remaps experts, model state, CPU/GPU work, and agent state reuse to the bandwidth and memory actually available on a personal machine. The authors report support for more than 20 MoE models, from 35B models on an 8GB laptop GPU to a 753B GLM model on one workstation GPU.

A POLICY ALGEBRA FOR TRUST-PRESERVING AGENTIC AI EXECUTION

The paper proposes a system for enforcing an AI agent's permissions throughout an entire task, not just when the task begins. For example, a refund agent could read the correct customer's record, calculate a refund, use the payment tool only below its spending limit, request human approval when necessary, and leave an audit trail, all under one combined set of rules. The authors report that their runtime stopped or corrected 94.8% of rule-breaking actions while still completing 86.9% of legitimate tasks.

NVIDIA'S AI MOAT IS SHIFTING FROM CHIPS TO CAPITAL

Nvidia is leveraging its capital strength to maintain AI leadership, moving beyond chips as competitors like AMD and Google gain ground. Through a $105 billion investment in an Ohio data center supporting OpenAI, Nvidia aims to sustain growth amid rising competition by expanding its AI infrastructure. By forming partnerships with Wall Street financiers for GPU investments, Nvidia seeks to widen its market influence and diversify revenue streams.

RETHINKING THE DATA MOAT

Dwarkesh Patel discussed AI research automation with Ryan Greenblatt, who emphasized algorithmic progress as a major driver of AI advancements, suggesting improvements even without human expert data. Shuchao Bi, co-founder of YouTube Shorts, echoed similar sentiments by highlighting the need for refined data distributions to enhance AI models. Both experts argue that AI research could accelerate with better data processes, challenging traditional notions about human data's significance in model improvements.

TRY PAVE FREE →

ANTHROPIC PREPARES SUPERVOTING POWER FOR FOUNDERS AHEAD OF IPO

Anthropic plans to give its co-founders a class of stock with extra voting power to help insulate them from external shareholder pressure.

BIRDS DON'T FLY LIKE PLANES. NEITHER DOES AI.

Local models, like Qwen3.8-27B, outperform larger cloud models, like GLM-5.2, despite being smaller because they rely on reasoning over memorization.

INTRODUCING HARVEY II

Harvey II enhances legal work efficiency by enabling AI agents to start with context, memory, and preferences from prior tasks, reducing repetitive setup.

VERCEL'S $1 MILLION SANDBOX ESCAPE CHALLENGE

Vercel launched a two-week security challenge offering up to $1 million to researchers who could escape its Firecracker-based Sandbox.

FROM ZERO TO ONE

Etched shipped its first hardware rack to Jane Street and raised $700M at a $21B valuation.

WARP'S NEW SYSTEM IS AN OUT-OF-THE-BOX SOFTWARE FACTORY FOR AI DEVELOPMENT

Warp Factories is an out-of-the-box system for streamlining AI software development using a software factory model.

CITI EXPECTS TO LAUNCH BITCOIN CUSTODY LATER THIS YEAR

Citi confirmed its planned digital asset custody service, first disclosed in November 2025, will launch later this year starting with bitcoin, letting clients access traditional and crypto custody in one framework. It's part of Custody+, a new Citi Investor Services platform built for continuous trading and shorter settlement cycles, also offering real-time asset servicing, instant settlement, liquidity tools, and AI-powered market intelligence. Citi is separately working with ICE on tokenized securities, joined a Swift tokenized-deposit pilot in July, and is part of a Clearing House-led tokenized deposit network targeting a first-half 2027 launch.

TRUMP ATTENDS WHITE HOUSE CRYPTO SUMMIT WITH SEC AND CFTC HEADS

Trump will personally attend Wednesday's White House crypto summit, joined by the heads of both the SEC and CFTC. Executives from Coinbase, Ripple, a16z, Chainlink, Paradigm, and Kalshi are expected, with the CFTC also hosting Coinbase, Robinhood, Kalshi, and Polymarket CEOs at its first Innovation Committee meeting Thursday. The meeting comes during a less-than-likely chance of CLARITY passing.

S&P 500 PERPS NOW LIVE ON COINBASE DERIVATIVES

Coinbase Derivatives launched a single-contract perpetual giving US traders perpetual-style exposure to all 500 companies in the S&P 500, the first such product available in the US. It enables leverage on the top 500 companies and allows users to take a strong bet on the short-term future of the US market.

STRIPE'S $7B OPENROUTER ACQUISITION: MICROPAYMENTS AT SCALE WITHOUT A BLOCKCHAIN

Stripe agreed to acquire AI model gateway OpenRouter for over $7 billion, gaining a platform that processes 200 trillion tokens monthly for 10 million users across 500-plus AI models and 80-plus model providers. OpenRouter's prepaid credit system handles AI inference micropayments as deposit balances rather than on-chain settlements, charging 5% on USDC top-ups versus 5.5% on card purchases with no markup on inference pricing. The deal expands Stripe's blockchain stack alongside stablecoin issuer Bridge, wallet provider Privy, and a Machine Payments Protocol co-development with Tempo, which settles at roughly one transaction per second. Cloudflare, Amazon, and Mastercard each launched on-chain AI agent payment rails in 2025, putting Stripe in direct competition across both custodial credit-balance and settlement-layer approaches to machine-to-machine payments.

CORRELATED PAIRS: HOW AMMS WIN THE BIGGEST MARKETS

Tokenization's primary structural shift is not settlement efficiency but the formation of correlated asset pairs (NVDA/SPY rather than NVDA/USD), which eliminate the hedging costs embedded in delta-neutral professional market making and let passive AMM strategies close the performance gap against firms like Citadel Securities, which earned $12.2B in net trading revenue last year. The mechanism mirrors DeFi's existing architecture: passive LPs supply correlated pairs where inventory risk is low, while active LPs compete over a small number of high-volume bridge pairs (the SPY/USD equivalent of ETH/USDC), concentrating professional attention where it earns sufficient return. Early evidence supports the thesis: ten tokenized equities trading against SPY in Uniswap pools on Robinhood Chain generated $33M in volume from 11,000+ traders in their first twelve days, including direct stock-to-stock swaps that bypassed dollars entirely. The passive vehicle that made no active decisions now beats roughly 90% of active managers. Passive AMM liquidity is expected to follow the same path as equity markets migrate on-chain and reorganize around correlated pairs.

WHY AI AGENTS NEED CONSTRAINED WALLET PERMISSIONS

Since crypto markets never close, and transactions are irreversible, AI agents with wallet access need traditional-finance-style controls (transaction limits, multisig approval, hardware security modules, delayed settlement, address whitelists, and contract simulation) rather than one broad permission set. AI agents reportedly settled only about $73 million on-chain between May 2025 and April 2026, versus roughly $46 trillion in stablecoin settlement over the same period. This gives the industry a window to build controls before agent-managed capital scales up.

RUSSIA MOVES TO RESTRICT RETAIL CRYPTO TRADING TO BITCOIN, ETHER AND USDT

Russia's central bank will restrict retail crypto trading to Bitcoin, Ethereum, and USDT starting September 1, implementing legislation that created the country's first regulated trading framework. USDT is the sole approved stablecoin, excluding USDC, DAI, and other alternatives from regulated channels. Non-qualified investors face a 300,000 ruble (~$3,600) annual purchase cap per intermediary, while qualified investors have no ceiling, though the per-platform scope means aggregate exposure scales with the number of brokers accessed. Crypto payments within Russia remain prohibited under current law, drawing a firm line between regulated investment activity and transactional use.

JUSTIN SUN'S HTX "DUSTING" CAMPAIGN AGAINST COINBASE AND BINANCE

Justin Sun is sending small USDT amounts from HTX (under EU/UK sanctions since May 26) to addresses that previously interacted with Coinbase, Binance, and Hyperliquid, triggering Coinbase account freezes for recipients until they prove they weren't dealing with a blacklisted entity. The move comes as Binance prepares to restrict deposits/withdrawals involving HTX and ten other platforms from August 23, and is read as pressure to force Coinbase, Binance, and EU regulators to either legitimize HTX or face a backlash from affected users.

BITPANDA BECOMES FIRST CRYPTO FIRM PUBLICLY FINED UNDER MICA IN AUSTRIA

Austria's regulator fined Bitpanda €70,000 for missing white-paper deadlines and publishing non-compliant marketing, marking the country's first MiCA enforcement action since the rules took effect in 2024.

WEEKLY CRYPTO CARD INDUSTRY DIGEST, AUG 10-16

EtherFi's next-gen app went live with five collateral types and a dedicated Aave V4 instance on Optimism powering its credit backend, RedotPay launched stablecoin loans against BTC/ETH plus three new payout currencies, Wirex launched its first BaaS partner on Arc, and roughly 15 other card products shipped features across the week.

PUMP.FUN CO-FOUNDER SAYS HE DOESN'T BELIEVE IN DECENTRALIZATION

Noah Tweedale told Crypto Insider he's a massive bear on decentralization, arguing internet history shows full-stack companies with strong UX win over decentralized infrastructure.

BITCOIN HOLDS $64K AS 30-YEAR TREASURY YIELD HITS 5.33%, HIGHEST SINCE 2007

The 30-year US Treasury yield hit 5.33%, its highest since 2007, as a global bond sell-off pushed UK 30-year gilt yields to 5.834% and Japan's 30-year JGB to 4.126%, while Brent crude topped $91 on US-Iran escalation.