This week the agent wars moved from capability to price, ChatGPT started acting on its own, WhatsApp said it will charge for business messages again, OpenAI offered Washington a piece of the company, and Alibaba banned Claude Code. Five moves that set the direction of 2026.
I’ve filtered out the noise for you
The pattern this week is control and cost. The labs are racing to make autonomous agents cheap enough to run everywhere, the platforms are moving to charge for the access you used to get for free, and governments are becoming players in the AI business, not just referees. Here is what matters for your team.
1. Anthropic Launches Claude Sonnet 5: Cheaper, More Powerful Agents
Anthropic launched Claude Sonnet 5 on June 30 with gains in reasoning, tool use, and coding, at introductory pricing of $2 per million input tokens and $10 per million output tokens through August 31. It can plan, use browsers and terminals, and run autonomously at a level that used to require bigger, costlier models.
Why it matters: Cheaper, capable agents put serious sales and operations automation within reach of teams without a huge budget. The agent race is shifting from raw capability to price per token.
2. ChatGPT Can Now Run Scheduled Tasks on Its Own
ChatGPT rolled out scheduled tasks, letting you create, pause, edit, and delete recurring or timed actions from a new Scheduled sidebar. Tasks can run up to once an hour, search the web, check connected apps, and notify you only when something changes. It is available on web and mobile for Go, Plus, Pro, Business, and Enterprise plans.
Why it matters: Routines like daily email summaries, weekly reports, or follow-up reminders now run on their own, no manual chatting required. It is a real step from an assistant that answers to an assistant that acts.
3. WhatsApp Will Start Charging for Business Messages Again
Meta announced that starting October 1 it will charge for utility and service messages on the WhatsApp Business Platform that were previously free, and starting August 1 it will bill AI agent messages by token at $2 per million tokens. One AI message can use 20,000 to 25,000 tokens.
Why it matters: In Latin America, WhatsApp is the main sales and support channel. Charging for messages that used to be free changes the economics of every conversation, and reminds everyone the channel is rented, not owned.
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. OpenAI Offers the US Government a 5% Stake
The Financial Times reports OpenAI is in early talks to hand the US government a 5% stake, worth roughly $42.6 billion on an $852 billion valuation. Sam Altman pitched the idea directly to President Trump, modeled on the Alaska Permanent Fund.
Why it matters: A government-company alliance, instead of hard regulation or abrupt bans, lowers the risk of sudden restrictions. Altman is not giving away shares, he is buying political protection with equity.
5. Alibaba Bans Employees from Using Claude Code
Alibaba banned employees from using Anthropic’s Claude Code starting July 10, after researchers flagged hidden code that could identify whether a user was based in China or tied to Chinese AI labs. Alibaba added Claude Code to its list of high-risk software.
Why it matters: Geopolitics now decides which AI tools you can use depending on where you operate. AI is no longer just software, it is contested territory.
Bonus: Content of the Week
Coinbase shared that it cut its AI costs by 50% while keeping the same token usage, by switching to open-source models. Expect more companies to chase the same savings.
