Got this newsletter from a friend? Subscribe for free →

Happy Tuesday! Anthropic wanted us to start the week off with a bang, so it did the only thing it could do: It dropped its updated S-1/A filing.

Not only did we learn that Anthropic is apparently very close to developing a model capable of Recursive Self-Improvement (RSI)—a model that can teach and improve itself—but we were also informed that, unsurprisingly, Anthropic spent a whole lot of money in Q3.

Let's get started.

— Jody Serrano

→ Anthropic plans to spend a lot of money over the next few years
→ Voting with your AI chatbot
→ Grok is costing Elon too much money

...and all the AI news we couldn’t fit here.

The news: In updated IPO documents filed on Monday, Anthropic revealed a massive increase in revenue year over year, jumping to $17 billion in Q3 2026 from $1.4 billion in Q3 2025.

Those weren't the only jaw-dropping numbers in the documents, though the others don't exactly elicit joy. The AI company reported a $23.3 billion net loss in the same quarter as its impressive revenue growth, nearly double the loss from a year earlier.

Anthropic also shed light on its plans for long-term spending, which was the biggest number of all. The company reported $895 billion in financial commitments, primarily for compute power for its AI models. About half of its total commitments are due by 2030.

The takeaway: Anthropic is clearly aware of the market's growing unease around AI spending and wanted to release its financial information in controlled doses to avoid spooking investors. It makes sense that it would release information that was more positive, such as its $112.4 billion in cash, first.

The news: Don't expect your AI chatbot to tell you how to vote. In fact, Gemini, Google's chatbot, doesn't even give you a straight answer when you ask it who won the 2024 presidential election. (Google says this is due to a technical problem that it's working on fixing.)

Experts say chatbots don't tell people who to vote for because they don't want to be accused of shaping the outcome of an election. However, while they may not be blatantly for or against any candidate, they may still impact the election in other ways.

"It’s not [that] AI is going to change the election because it’s telling you to vote a certain way," Jason Thatcher, a business professor at the University of Colorado Boulder, told Gizmodo. "It’s going to shape how you think about the candidates. It’s going to shape how you think about the issues. It’s going to shape how pervasive you think a problem is and how deep a problem is."

The takeaway: It's still not clear whether chatbots can influence opinions on political topics or voting. Given that only 8% of American adults are using chatbots to find information on elections in 2026, we may not have the answer to that question for some time.

The news: Elon Musk was forced to do an about-face earlier this week after saying that, "Saving money for the distant future is pointless with the advent of super intelligent abundance."

Three days later, Musk quietly walked back his proclamation. The shift occurred when users on X complained about a new token limit on Grok Bot, an always-on AI agent developed to compete with OpenAI's Dots and Meta's Muse.

"The cost of serving Grok Bot was unsustainably high relative to revenue," Musk said in a post on X.

The takeaway: It's still unclear whether AI will provide the vast financial wealth its proponents have promised. While companies like Anthropic are certainly reporting eye-popping revenues, their costs are even steeper. Some, like Oracle and Meta, are already looking for even more money.

The AI company warned that it is close to developing a model capable of Recursive Self-Improvement (RSI), or training itself to improve.
Read more →

Oracle Looks to Raise $20 Billion to Fund AI Ambitions

The Shopping Bots Are Coming for Amazon

Anthropic warns that it’s very close to RSI capability, aka self-building AI

Chatbots Are Leading Hikers Astray, With Dangerous Consequences

An AI Agent Injected Drugs Into a Live Mouse’s Brain

Hundreds of Mundane AI Apps Are Hiding Sexually Abusive Deepfake Features

Dartmouth’s provost warned us about AI cheating—but his own AI use cost him his job

Anthropic's colossal pay packages cost the company more than $100,000 a month per employee

OpenAI says its AI is getting better at doing things humans do, like shopping and meal planning

OpenAI is finally merging ChatGPT’s Chat and Work tabs