Tech News
Live coverage of AI models, platforms and the companies shaping them, curated by our newsroom and updated as stories break.
Top Story
07 Aug 2026 · AI ModelFive companies that normally fight over the same developers agreed on something on August 6, 2026: one way to package the add-ons that extend AI agents. They call it Agent Plugins 1.0.0, an open standard from OpenAI, AWS, Microsoft, Cursor, and Vercel. The pitch is simple. Build a plugin once and it runs in any tool that supports the format, instead of rebuilding it for each one. The name missing from the list is the part worth noticing: Anthropic, which created both of the technologies the new format bundles together.
Five companies that normally fight over the same developers agreed on something on August 6, 2026: one way to package the add-ons that extend AI agents. They call it Agent Plugins 1.0.0, an open standard from OpenAI, AWS, Microsoft, Cursor, and Vercel. The pitch is simple. Build a plugin once and it runs in any tool that supports the format, instead of rebuilding it for each one. The name missing from the list is the part worth noticing: Anthropic, which created both of the technologies the new format bundles together.
When Claude Fable 5 launched on June 9, its biology filters were tight enough that people kept getting bounced for ordinary questions. Asking it to read a blood test result, or explain a symptom, or help study for a cell biology exam often got quietly handed off to a weaker model, and users noticed fast. On August 7, 2026, Anthropic rewrote the rules. It says the change cut those wrongful refusals by about 85% in its own testing.
A wave of free tutorials, resource guides, and full courses is teaching developers how to build custom automations and agentic workflows with Anthropic's Claude Code. The common thread across them is a change in how people use the tool: less back-and-forth chatting, and more reusable setups that Claude can run on its own. The material ranges from step-by-step walkthroughs for building a skill, to a curated list of dozens of tools, to a set of free certified courses from Anthropic Academy. Together they are pushing Claude Code from a handy assistant toward a core part of the development workflow.
Developers are flipping the usual advice about AI. Instead of chasing the perfect prompt, they let the model interview them first. The technique, called reverse prompting, is simple: dump everything about your goals into the AI, then ask it what else it needs to know before it suggests a plan. It is spreading alongside a second idea
Developers are trading tricks to stop Anthropic's Claude from answering simple questions with piles of unrequested code. The common gripe is that Claude is strong at coding but too eager with it, often shipping a full implementation even when someone asks something basic like "How does this algorithm work?"
Anthropic's most capable publicly available AI model went dark on June 12, three days after launch. The US government ordered it offline over national security concerns. A jailbreak was involved, and Amazon played a role most people did not expect. Developers who had spent the week building on it woke up to find it gone, with no advance warning and no clear timeline for return. For 19 days, the most powerful AI model available to the public simply did not exist.
