It is terribly easy to poison a large language model (LLM). What’s even more annoying is that you don’t need to have access to it to do this.
A new study from NYU Scientists are deeply studying how much medical misinformation an AI can handle before it starts giving wrong answers. They found that even a tiny amount of false information – just 0.001% of the data it learns from – can disrupt AI.
While the study focuses on someone deliberately trying to trick the AI during training, it also highlights a bigger problem: all the misinformation already circulating online that these AIs are learning from. Additionally, there is the problem of outdated medical information still lingering in some databases. Sci-Hub, a free access/piracy site for scientific research, found that only 15% of articles removed by their editor were removed from the site.
The good news is that any poisoned data must compete with what might be accurate information for attention. Thus, the ability to poison an LLM may depend on the subject. The bad news: LLMs are trained on data scraped from the internet, and much of it is garbage.
Now here is this week’s roundup of AI-powered martech releases:
Amazon Ads announced a new feature within Amazon Marketing Cloud (AMC). This new feature allows advertisers to generate SQL queries for the desired audience using natural language. This eliminates the need to write code, significantly reducing the time required to develop audience queries. Advertisers can then run these queries in AMC to create new audiences and activate them on Amazon DSP and the advertising console.
CordialCordial Edge uses multimodal AI to generate personalized marketing models for individual retail brands. These models take into account various types of data, including brand creatives, illustrations, photographs and text, to optimize marketing messages to increase purchases.
PearlMountain Limited has released an update to its video creation platform, FlexClip. This update includes two new tools: an AI video generator and a motion tracking feature. The AI Video Generator enhances creative possibilities, while Motion Tracking simplifies complex editing tasks.
Jellypod launched its AI podcast studio. This platform offers features like AI-based voice cloning, content control, and robust content integration. Jellypod aims to simplify podcasting by helping creators with all aspects of production, from ideation to final edits and publishing.
More dapper announced the availability of AskAI banner advertising deployment. This feature allows digital publishers to integrate an AI-powered conversational response engine into banner ads. This in-ad unit provides contextual awareness of user content and offers advertisers a high-attention environment to connect with consumers.
HUMAN Safety in partnership with TollBit to protect against unauthorized AI scraping bots, enforce robots.txt guidelines, and enable publishers to monetize their content.
Jivox Updated Jivox IQ DaVinci platform. This update includes deep API integrations with leading retailer DSPs, such as Amazon DSP and Walmart Connect DSP. The platform combines first-party audiences with creatives and media into a single workflow.
Complete advertising science launched its AI-powered Total Media Performance (TMP) solution. TMP integrates media quality signals with ongoing optimization to help advertisers achieve better campaign results.
Creative realities launched AdLogic CPM+TM, a campaign planning and management platform. This platform offers programmatic capabilities and is designed for in-store retail media networks. AdLogic CPM+TM allows advertisers and publishers to plan, manage and optimize their campaigns within a unified AdTech stack.
BrandRep launched an AI-powered keyword generator. This tool uses artificial intelligence to identify relevant keywords for small and medium-sized businesses. The goal is to improve SEO performance and drive growth in competitive digital spaces.
Add MarTech to your Google News feed. 
About the author
Constantine von Hoffman is editor-in-chief of MarTech. A veteran journalist, Con has covered business, finance, marketing and technology for CBSNews.com, Brandweek, CMO and Inc. He was an editor at the Boston Herald, a news producer at NPR and has written for Harvard Business Review , Boston. Magazine, Sierra and many other publications. He has also been a professional stand-up comedian, has lectured at anime and video game conventions on everything from My Neighbor Totoro to the history of dice and board games, and is the author of the realistic novel and magical John Henry the Revelator. He lives in Boston with his wife, Jennifer, and too many or too few dogs.
