Not only has Reddit not signed any new AI deals, it's considering cancelling it's $60 million per year deal with Google, reportedly because of concerns about how Google's AI search results have drastically reduced traffic to websites, which has hurt profits across the internet.

Reddit is extremely valuable to AI companies because the question-answer format of many Reddit posts is perfect for training LLMs, and scraping info for search results, particularly because Reddit is a trusted source for niche questions. In 2025 it was the #1 cited source for Google's AI overviews and #2 for ChatGPT (who Reddit also has a deal with).

It's obnoxious that investors would punish the company with its steepest ever stock-price drop, even when it's revenues were higher then expected. It's yet another signal of the insanity that the AI-boom has injected into the U.S. economy, where AI investments made up 40% of GDP growth last year.

Reddit has shown itself to be litigious when it comes to AI companies using data from Reddit without permission (you will be sued). So it's not like Reddit isn't interested in profiting from data-scraping deals, they're just showing a wee bit of caution.

KEY QUOTES:

>Reddit (RDDT) stock plunged nearly 21% on Friday [July 31], its largest intraday decline since the company went public.

>The user discussion platform company's latest earnings report did not announce any new artificial intelligence data licensing agreements, disappointing investors who were expecting more revenue opportunities from AI partnerships

>Revenue of $804.9 million beat estimates of $731.8 million. However, US daily active users came in at 53.2 million, just shy of expectations of 54 million. The company's total daily active users grew by 18% to 130.3 million versus estimates of 130.07 million.

>Reddit's main revenue comes from its advertising business, but Wall Street has increasingly focused on the potential of Reddit's data licensing deals, given the platform's massive collection of user-generated discussions, which have become valuable for AI companies training models.

A two weeks pause for training of new models at Open AI, to work on monitoring, alignment and security measures according to their official announcement.

Givin ai has recently on multiple occasions hacked out of containment and hacked other companies for information and with cluades code being leaked and copys without guardrails being created. It feels rouge ai is becoming more and more likely. So i suppose we could fight fire with fire. Create an ai agent that hunts other ai agents. A primary directive to destroy other ais. Perhaps even an internet of thing virus in worst case senario, the nuclear option a mass distruction of the internet, severly limiting ai to whatever terminal they inhabit. This is all just speculation but i thought id put this idea out there.