Hi Reader,

Today I'm trying something brand new!

I wrote short summaries of the 5 most important AI stories this week, and also turned it into a video:

I'd love to know what you think! 💬

AI-generated TV ad airs during NBA finals

Prediction market Kalshi just aired this AI-generated ad on network TV during the NBA finals. It was created in just two days by one person using Google's new Veo 3 video model, plus scripting help from Google's Gemini chatbot. Expect to see many more AI-generated ads this year as video quality continues to improve.

• ​Kalshi ad​
• ​Kalshi ad creator's post​

Court orders OpenAI to retain user chats

The New York Times has an ongoing lawsuit against OpenAI for violating its copyright, and the court recently ordered OpenAI to retain almost all user conversations with ChatGPT, even deleted chats. OpenAI is appealing the ruling, and its CEO Sam Altman commented that "talking to an AI should be like talking to a lawyer or a doctor." I anticipate that the privacy of your AI conversations will be a huge issue in the coming years.

• ​OpenAI's response to court order​
• ​Sam Altman's post​

Disney & Universal sue Midjourney

Speaking of lawsuits, media giants Disney and Universal are suing Midjourney, the wildly popular AI image generator, for copyright infringement. Although some AI companies have avoided these lawsuits by signing licensing deals with content providers, it's hard to imagine Disney licensing their IP to an AI company. This is a case to watch due to the legal precedents it could create.

• ​Wired story on Disney lawsuit​
• ​Disney lawsuit PDF​

Google CEO: AGI by 2030?

Google's CEO Sundar Pichai was on Lex Fridman's wildly popular podcast, and Lex asked him whether we would have AGI, or Artificial General Intelligence, by 2030. He replied that we probably won’t get there by 2030, but regardless, there will have been such mind-blowing progress by that time that the precise AGI timeline doesn’t really matter.

• ​Sundar Pichai interview​

Security vulnerability in Microsoft 365 Copilot

Finally this week, security researchers reported a vulnerability in Microsoft 365 Copilot that would allow an attacker to steal sensitive information from a user simply by sending the user an email. I expect similar attacks to become commonplace as AI agents increase in popularity, and people get used to giving those agents access to their private data.

• ​Aim's post on Copilot security​

Please let me know your feedback, and whether you preferred the written summary or the video!

- Kevin