Welcome back to Daily Zaps, your regularly-scheduled dose of AI news ⚡️ 

Here’s what we got for ya today:

• FTC launches probe into AI chatbot risks

• Presidential AI Challenge

• Broadcom to build chips with OpenAI

• Anthropic to stop selling to China

Let’s get right into it!

Sign Up | Advertise | Tools

GOVERNMENT

FTC launches probe into AI chatbot risks

The U.S. Federal Trade Commission is preparing a major study into the risks posed by AI-powered chatbots from companies like OpenAI, Google, and Meta, focusing on privacy, data handling, and potential harms to children and other users. Using its 6(b) authority, the FTC will compel the nine largest chatbot providers to share information, reflecting growing scrutiny despite White House guidance urging lighter regulation to encourage AI growth.

The review comes amid lawsuits, including one against OpenAI tied to a student suicide, and rising concern about chatbots influencing vulnerable users with harmful content. While tech CEOs meet with the White House to discuss AI adoption in schools, the FTC’s probe underscores the tension between promoting innovation and safeguarding public safety.

EDUCATION

Presidential AI Challenge

Top U.S. tech leaders, including Google’s Sundar Pichai, IBM’s Arvind Krishna, and OpenAI’s Sam Altman, joined government officials in Washington to launch the “Presidential AI Challenge,” a new initiative focused on bringing artificial intelligence into K-12 education. The event emphasized AI’s role as a powerful tool for preparing students for the future, with leaders highlighting its potential to transform classrooms, enhance learning, and inspire the next generation of innovators.

Education Secretary Linda McMahon underscored that AI should be embraced rather than feared, while Pichai praised the effort as a way to encourage students nationwide to use technology in extraordinary ways. The initiative aligns with broader federal efforts to position the U.S. as a global leader in AI, combining executive action with industry collaboration to accelerate safe, responsible, and wide-scale adoption of AI in schools.

BIG TECH

Broadcom to build chips with OpenAI

OpenAI will produce its first custom AI chip next year in partnership with Broadcom, aiming to meet surging compute demand and reduce dependence on Nvidia. Broadcom’s CEO Hock Tan hinted at a $10 billion commitment from a new client, later confirmed by insiders to be OpenAI, which plans to use the chips internally rather than sell them.

The move follows Google, Amazon, and Meta, which also design specialized AI hardware, and marks a major boost for Broadcom’s custom chip business, lifting its growth outlook and share price. While Nvidia remains dominant, its growth has slowed, and analysts expect Broadcom’s custom chips to gain market share by 2026. OpenAI CEO Sam Altman has emphasized the urgent need for more compute to support products like ChatGPT and new models such as GPT-5, with plans to double its compute capacity within five months.

STARTUPS

Anthropic to stop selling to China

Anthropic is tightening its Terms of Service to block not only users in restricted regions like China but also companies controlled by entities in those jurisdictions, even if they operate abroad. The company warns that authoritarian governments can compel businesses to share data or aid intelligence efforts, creating national security risks when such entities access advanced AI.

By prohibiting subsidiaries and organizations more than 50% owned by firms in unsupported regions, Anthropic aims to prevent misuse of its models for adversarial military, intelligence, or competitive AI development. The update aligns with its broader push for strong export controls, domestic infrastructure for AI scaling, and rigorous national security evaluations, emphasizing that responsible AI development must advance democratic interests and guard against exploitation by authoritarian adversaries.