OpenAI's Agents API is now in public beta. The API brings the same harness and infrastructure that powers Codex to developers through a simple and flexible API. It allows developers to create useful agents with a powerful harness that manages context, uses tools efficiently, and coordinates subagents. The API provides infrastructure to keep agents running reliably for days, and environments where agents can work with files, run code, and save intermediate results.

Anthropic claims it has disrupted several plots to use its AI models to conduct research that could have helped develop biological weapons. The company says it still can't determine whether a query serves a legitimate or nefarious purpose, so it erred on the side of caution as the consequences of mistakes could be severe. Anthropic detected use from companies from blocked regions and parties that otherwise tried to obfuscate their purpose.

Unitree CEO Wang Xingxing recently unveiled details about his company's IPO in an interview that covers his business principles, Unitree's managerial structure, and how the company is changing its employees' lives. Unitree listed on the STAR Market on August 19, with its market capitalization briefly reaching 440 billion yuan after its IPO. Wang was briefly the richest person in China born in the 1990s.

Analysts predict that AI will soon deliver double-digit growth in advanced economies. Experts say it could push growth to something like 10% to 15% per year. The decade ahead could see economic growth of 30% per year and beyond. Predictions of double-digit growth are common among AI insiders.

Companies are experimenting with cheaper open models on inference providers. There is a trend toward more investment in model routing to route simpler requests to cheaper models. The industry is seeing more knowledge-sharing sessions on how to use AI models cost-effectively. More and more companies are setting per-developer limits on AI usage.

This post follows the journey of a PostgreSQL query through all of the layers of a complex sharded system. It aims to help readers understand what goes into making a large-scale sharded PostgreSQL deployment appear to be a single server even when spanning thousands of servers. A real database has hundreds of tables - the example in the post keeps the schema simple with two tables spread across four shards.

Centralized router platforms can offer customers better economics than local deployments. Outcomes pricing belongs to the infrastructure layer. The model layer could offer outcomes pricing, but it will have the same problems as the app layer. Router layer pricing is most politically aligned with customers, who just want to sell more compute and don't care which model is used.

The 'next thing' is always simpler than you think and very obvious in hindsight. OpenAI's addition of Computer Use is deceptively simple and obvious in hindsight. 'Computer use' is a very generic, unspecified thing. It requires general intelligence across many domains to get better at it.

Your vendor says "OWASP LLM Top 10 covered." That's a checklist, not a defense. Seven red flags every developer should catch before signing.

Astra will change the world one way or another.

Google has six months to implement the technical measures and finalize the processes for data access.

The people building AI earnestly believe that it could kill us all by the end of the decade.

AI can deidentify you better than many adversarial bot programs.

Most changes that would have made agents better at codebases make agents better too.

A storage solution is only part of what a team needs to collaborate around code.