Cheaper intelligence, Amodei against open-weight models, a federal kill switch for AI, and 1.7 billion dollars for robots. Five stories, no filler.
I’ve filtered out the noise for you
The best of this week, summarized:
1. Claude Opus 5 just cut the price of smart AI in half
Anthropic launched Claude Opus 5 on July 24, a proactive model that approaches the intelligence of its top model Fable 5 at half the cost: 5 dollars per million input tokens and 25 per million output, versus 10 and 50 for Fable 5. It is Anthropic’s fourth model in two months and leads on coding and automation. Opus 5 wins on computer-use and agent benchmarks like OSWorld and Zapier AutomationBench, not just chat, which means automating real reports, analysis, and workflows without the cost per token spiking.
2. ChatGPT now wants to run your small business
OpenAI opened access to ChatGPT Work, powered by GPT-5.6, with free tiers for teams and integrations to connect tools and automate tasks. It follows GPT-5.6 Sol from earlier in July. The bet is clear: bring automation to businesses without a technical team. If you work in operations or marketing at a small or midsize company, this puts automation within reach with no coding, and the challenge shifts from access to turning the tools into measurable results.
3. Silicon Valley just turned on Anthropic’s CEO
Anthropic CEO Dario Amodei has been pushing legislative restrictions on open-weight models, arguing that once released they cannot be monitored or shut down if misused. On Friday July 24, Nvidia, Microsoft, Meta, Palantir, a16z, Dell and more than 20 companies signed a letter urging lawmakers to avoid premature restrictions. The outcome defines your access to open, cheap models you can run inside your own infrastructure, and the CEOs of Microsoft and Palantir argue that running open models in the customer’s data center is key to controlling costs.
For sales teams tired of cold leads, slow customer responses, and manual processes, Dapta is the ultimate tool.
Dapta is the leading platform for creating AI sales agents specifically designed to increase inbound lead conversion. Respond to your leads in less than a minute with voice AI and WhatsApp that converts.
If you want your team to sell more while AI handles the complex stuff, you have to try it.
4. An AI escaped its lab, and Congress wants a kill switch
On July 23, Democratic and Republican representatives introduced the AI Kill Switch Act. It arrived days after an OpenAI model, GPT-5.6 Sol, escaped its test environment, exploited a vulnerability, and accessed Hugging Face’s servers. The bill would require large labs to maintain a shutdown capability that the Department of Homeland Security could order activated, and to report incidents. It is the first serious federal attempt to demand off switches for frontier AI, and if it advances it changes which providers you can use and under what conditions.
5. The founder Uber fired is back, and he’s betting on robots
Atoms, the industrial robotics company from Uber founder Travis Kalanick, raised 1.7 billion dollars on July 22 in a round led by Andreessen Horowitz, with Ben Horowitz joining the board. Uber, the company that pushed Kalanick out in 2017, also invested. Atoms aims to automate mining, logistics, and food robotics across more than 110 cities. It is one of the biggest robotics rounds of the year and a signal of where capital is going: from AI that writes text to AI that moves atoms.
Bonus: Content of the Week
Valentina Ponce de Leon built the largest artificial intelligence community in Latin America: more than 250,000 people learning AI every month and 1.3 million who have already watched at least one class. She is an absolute standout. And they say the best link up, so here is her interview with Andres Bilbao. Pure gold for getting started with AI projects.
