Tech blogger Simon Willison has written a 3-blog series about the Prompt Injection vulnerability which Preamble discovered. Contact Preamble at sales@preamble.com for help with defending against this attack.

Preamble discovered a critical safety issue in large language models, which we call Prompt Injection, and the discovery has garnered significant media attention.

The Preamble team was in Cannes April 14-16th for the first annual World Artificial Intelligence Cannes Festival (WAICF) on all things AI. The conference brought together top AI researchers and companies from all over the world.

A new bill was introduced to reduce and prevent online harm to children. Read how Preamble can help.

Internet safety for children is a significant concern at Preamble. We are doing our part to help platforms with a high-impact area to make safer decisions. 

Read an interview about women leading the AI industry from our Lead Engineer, Leyla Hujer.

The Preamble team was in Amsterdam for the 15th ACM RecSys Conference. We met a lot of talented researchers and industry professionals. We had many intriguing conversations, learned a lot, and saw a few common themes being discussed: regulating recommenders, key problems with recommendation systems, ensuring fairness, reducing bias, and how to better manage large recommendation systems. Read about our time at the conference.

On October 5, 2021, testimony from Facebook whistleblower Frances Haugen was presented to a Senate Subcommittee on Consumer Protection, Product Safety, and Data Security. Read our summary and how Preamble is working to mitigate these issues in the future.

Read the interview with our CEO on building a team and his experience as a second time founder.

Read our Chief Scientist's take on the role platforms play in creating content from curation and the current state of laws that protect these platforms.

The authors describe why middleware would be a good solution for internet platforms to adopt where users would have more control over the content that they see. They also describe how a middleware service protects the user's First Amendment Rights.

Online platforms currently use various content moderations solutions to remove content that should not be on the platforms, but there is no solution to say what content should be shown. In this case, the author is describing content recommender systems that are the basis of modern media infrastructure and several promising improvements that can be made.

We, humans, have a notoriously difficult time specifying what we actually want, and the AI systems we build suffer from it. With AI, this warning actually becomes “Be careful what you reward!”.