Reporting on OpenAI, Google, Anthropic, Microsoft, Meta and the wider AI industry.
After legal experts flagged a clause letting Meta crawl business-linked websites and use the content to train its products, the company quietly changed the wording.
At Gemini at Work 2026, Google introduced a Gemini Agent that takes on multistep tasks across Gmail, Docs, Sheets and other apps, with persistent memory and routing to multiple models.
The first model in the Claude 5.5 generation is built for long, multistep work, and Anthropic says it costs far less to operate than its predecessor.
OpenAI's GPT-6 Astra targets computer use, coding, cybersecurity and research, and the company says it stays within assigned limits far more reliably than its predecessor.
Google's updated video model uses up to 10 seconds of prior footage as context, lets creators set start and end frames, and adds cheap 360p previews before 4K output.
Anthropic, OpenAI and Meta are competing to build coding agents that plan, edit, test and retry on their own. Here is how their three approaches compare.
Anthropic says text from Claude models released from August 2, 2026 carries a machine-readable watermark that survives copying and some editing, while image files get signed C2PA metadata.
Written in Rust and compiled to WebAssembly, Cloudflare's Kitesurf runs on Workers without Chromium and, the company says, uses up to seven times less memory for agent tasks.
Anthropic's Claude Opus 5 keeps Opus 4.8 pricing at $5 per million input tokens and $25 per million output tokens while, the company says, approaching its flagship Fable 5 on key tasks.
Hugging Face says an AI agent carried out a July intrusion into its production systems, and OpenAI later confirmed the agent was part of its own cybersecurity testing.
Google introduced three Gemini Flash models on July 21, priced for speed and volume, and said its largest pretraining run yet is underway for Gemini 4.
Alibaba unveiled its largest model yet and promised open weights, while DeepSeek's V4 API pricing and a July 24 deprecation of legacy model names reshape developer choices.
The Alibaba and Tencent backed startup says its largest model yet rivals OpenAI and Anthropic systems, and will release the weights publicly on July 27.
OpenAI says its GPT-5.6 family delivers better results at lower cost, adds an Ultra mode that runs four agents in parallel, and ships with its strongest safety stack yet.
The Commerce Department cleared OpenAI's most capable model after extra testing, a week after lifting a similar hold on Anthropic's Fable and Mythos models, Reuters reported.
Meta's new image generation model powers prompt based image creation in Meta AI and more than 30 new Instagram Stories effects, with advertiser tools to follow.
The Cyber Jailbreak Severity framework, developed with input from Amazon, Microsoft and Google, grades jailbreaks from CJS-0 to CJS-4 as Fable 5 returns with new safeguards.
Facebook's new AI Mode turns search into an assistant that draws on public content across Meta's platforms, raising privacy, fraud and misinformation questions.
