Prompt injections refer to a broad category of attacks on #LLMs and multimodal models—but what specific techniques are being used and how can we mitigate harmful effects?

In our recent blog post by Teresa Datta, you’ll find:
👉 An explanation of direct vs. indirect prompt

Really insightful podcast with our Chief Scientist, John P Dickerson! 🎧

For those of you attending Open Data Science East in Boston next week, John will also be giving a talk titled “All Models Great and Small: It’s 2024, You Don’t Have to Use GPT-4 for Everything Anymore” on Thursday,

Scenes from another incredible #TeamArthur offsite, this time in Puerto Rico! 📸💜🏝️

We’re grateful to have been able to unplug for a few days of team building, knowledge sharing, go-karting, horseback riding, ATVing, and so much more. Feeling motivated and excited for all

“Prompt injections” refer to a broad category of attacks on large language models and multimodal models meant to elicit unintended behavior.

In our latest blog post by Teresa Datta, we introduce a taxonomy to organize five key types of prompt injection attacks based on their

There’s no doubt that generative AI is causing a transformation across the business world—but many organizations are failing to implement guidelines that ensure the technology is enterprise-ready and used responsibly.

Join our webinar on Tuesday, co-hosted with CredoAI, to

You *could* use a chainsaw to simply cut a piece of paper... but wouldn’t scissors do the job even better? ✂️

Similarly, in many situations, it makes more sense to use a small language model over its larger counterpart.

Learn why bigger isn’t always better on the Arthur blog:

For those who missed last week’s Ground Truth event at our office, we were lucky enough to host a panel of five incredible, insightful women:
🟣 Tanya Dua of LinkedIn News
🟣 Teresa Datta of Arthur
🟣 Diana Hsieh of Reality Defender
🟣 Savannah Thais 👩🏼‍💻 of The Data Science Institute at Columbia University

💻 Don’t miss our webinar next week, co-hosted with CredoAI!

Arthur’s ML Research Scientist Teresa Datta and Credo’s Tech Policy Product Manager Ehrik Aldana will share how enterprises can ensure safe #generativeAI deployment by combining a governance wrapper and an #LLMOps tool

Looking forward to hosting this Wednesday’s panel with some of the most innovative women leaders in the field of AI!

Join us at Arthur HQ to hear from:
🔸 Cass Matthews (Microsoft)
🔸 Rebekah Tweed (@AllTechIsHuman)
🔸 Savannah Thais 👩🏼‍💻 (@datascicolumbia)
🔸 Tanya Dua (@LinkedInNews)

Check out these two exciting events on Wednesday with the Arthur team! 📆

👉 AWS Marketplace webinar with Arthur’s Chief Scientist John P Dickerson and VP of Engineering Zach Fry: pages.awscloud.com/awsmp-gim-1pck…

👉 Ground Truth event co-hosted with Work-Bench, with panelists from

We’re psyched to be celebrating Women’s History Month in a big way at Arthur HQ next week! 🎉

Join us on Wednesday as we learn from some of the most remarkable female leaders and innovators in the field of AI who are shaping policy, pioneering research, and building

Excited to have made the shortlist for the 2024 Cloud Security Awards in the “Best Firewall Security Solution” category! 👏

🛡️ Learn more about Arthur Shield, our LLM firewall that protects against hallucinations, sensitive data leakage, prompt injection, and more:

R&D has always been at the core of the Arthur identity, and we have always adopted a research-led approach to product development.

This experimental approach and Arthur’s expert researchers continue to drive exclusive capabilities in #LLMs , computer vision, bias mitigation, and

Hear from Arthur’s Chief Scientist John P Dickerson and VP of Engineering Zach Fry in our upcoming webinar with AWS Marketplace!

Attendees will learn how to:
👉 Understand evaluation criteria for both broad application needs as well as specialized use cases
👉 Evaluate models

#LLM leaderboards play a crucial role in advancing the field of NLP by:
👍 Fostering competition
👍 Encouraging model development
👍 Providing a standardized framework for language model performance comparison

But, it can be dangerous to take LLM leaderboards at face value. They

Happy #InternationalWomensDay to the incredible women of Arthur, of our community of customers and partners, and beyond!

Thank you for your unwavering commitment to innovation and excellence—and for continuing to drive change, inspire inclusion, and make a difference in the

Last week, at #AAAI2024 in Vancouver, Arthur presented recent joint work with Morgan Stanley at the Recommendations Ecosystem Workshop focused on Modeling, Optimization, and Incentive Design.

Check out our recent blog post written by ML Engineer Teresa Datta that recaps some

Successfully deploying #AI at scale comes with no shortage of challenges—here’s how we can help.

🚀 Launch
Arthur’s turnkey, plug-and-play solutions allow companies to build on top of their internal knowledge base and make informed, data-driven decisions when integrating the

Last call to subscribe before our newsletter is sent out today! 📧

Are you subscribed? 👉 arthur.ai/newsletter

At one of our recent Ground Truth events, Seek AI's ბეცალელი touched on how the “leaderboardization” of evaluations is often insufficient.

Read more about some of the potential dangers and downsides of #LLM leaderboards in our recent blog post: bit.ly/49F1JSK