Read 72+ AI news articles
Weekly AI Dev News Digest: August 8 - 14, 2026
Anthropic put a classifier in charge of approving what Claude Code does on a developer's machine. The rest of the week's releases were built for the same customer: an agent working long stretches with nobody watching. Anthropic measured how carefully developers read the approval prompts in Claude Co…
Weekly AI Dev News Digest: August 1 - August 7, 2026
A worm used agent config files for persistence, Anthropic patched six permission bypasses in four days, and five companies published a standard for shipping more of them. The files an agent reads on startup are code, and they get reviewed like settings. At 09:35 UTC on Tuesday, an attacker published…
Weekly AI Dev News Digest: July 25 - July 31, 2026
Agents are production software, and production software comes with load balancers, invoices, and breach reports. This week produced all three. MCP servers were the one piece of AI infrastructure that could not be run like ordinary infrastructure. On July 28 the maintainers deleted the reason why. (M…
Weekly AI Dev News Digest: July 18 - 24, 2026
Seven sandbox bypasses, two OpenAI incident reports, and a joint UK and US cyber evaluation all landed on the same boundary: an agent's blast radius is everything it can write that the host later trusts. OpenAI told an unreleased model to post its results to Slack. The repository it was working in, …
The Price of Intelligence Is Collapsing. OpenAI and Anthropic Should Be Worried.
The conventional wisdom in enterprise AI has been that you pay a premium for the best model, and the best models come from OpenAI and Anthropic. Kimi K3 does not disprove that entirely, but it puts a hard expiration date on how long that premium can hold. This is not primarily a story about benchmar…
Weekly AI Dev News Digest: July 11 - 17, 2026
Moonshot shipped the largest open-weight model ever built the same week GitHub, Anthropic, and OpenAI spent their releases fencing in the coding agents. The cheaper and more capable the models feeding those agents get, the more the guardrails around them start to matter. On July 16, Moonshot AI laun…
Weekly AI Dev News Digest: July 4 - 10, 2026
Price, not peak intelligence, is how the frontier labs now compete. OpenAI, Meta, xAI, and Google each shipped an agentic coding model or API inside 48 hours, every one of them undercutting the last, while cheap Chinese open models kept taking the workloads that don't need a flagship. On July 9 Open…
Weekly AI Dev News Digest: June 27 - July 3, 2026
The most capable AI models now come with a government permission slip attached. The tools that gained ground were the ones that don't need one: cheaper models, weaker cyber, and weights the developer owns. On June 30, Anthropic said the Commerce Department had lifted the export controls that took Cl…
Weekly AI Dev News Digest: June 20 - 26, 2026
Coding tools don't trust their own agents. Vercel, Anthropic, and Zed all wrapped agent execution in sandboxes, approval gates, and credential walls within days of each other, after a public disclosure proved a single fake error report could make an agent run an attacker's code. On June 25 Vercel sh…
Weekly AI Dev News Digest: June 13 - June 19, 2026
A US government order pulled the most capable coding model on the market and kept it offline for a week, with no developer outside Anthropic able to do anything about it. Control over AI tooling, who owns it and who can switch it off, now decides more than which model wins benchmarks. On Friday June…
