⛔ [Error] Trademark not granted

Was this email forwarded to you? Make sure to subscribe for more!

Greetings, fellow humans. 👋

This is Not A Bot - the newsletter about AI that was definitely not written by AI. I’m Haroon, CEO of Autoblocks and founder of AI For Anyone, and I share the latest news, tools, and resources from the AI space.

Some exciting news this week! We are proud to introduce AI Roundtable, a series that will explore the latest and greatest in the world of generative AI & discuss upcoming news, trends, and launches in the space. Check out some snippets here!

Now let’s dive into it…

Have a product, service, job, event, newsletter, app, book, movie, tool, or anything you'd like to share with over 40k subscribers? 

Advertise with us 

And since you’re here already, check us out on our socials - snacks on us if you also hit follow :)

Pallet Job board | Twitter | LinkedIn | YouTube

🧵 In today's edition:

• ⛔ [Error] Trademark not granted

• 🚧 NVIDIA Guardrails AI

• 🤗 HuggingFace announces chatbot

• 🤑 AI Fundraising News

🤖 Top AI News

⛔ [Error] Trademark not granted

OpenAI has been on a meteoric rise since it launched ChatGPT in November last year, giving the world its first taste of user-centric AI.

And as they would have imagined, an absolute blitz of GPT clones have since been announced:

ThreatGPT, MedicalGPT, DateGPT, DirtyGPT, TruthGPT… the list goes on.

In an attempt to slow things down, OpenAI applied to trademark the "GPT" in ‘ChatGPT’ last December, but the USPTO denied the company's request last week.

The reason? According to USPTO, OpenAI’s attorneys neglected to pay the associated fees and provide “appropriate documentary evidence supporting the justification of special action.”

What this means: The trademarking process could now force OpenAI to wait another five months, which could encourage the company's rivals to churn out GPT clones.

Altman will be sweating to get his ™ to try and slow down the GPT train or, at the very least, make some money off it. 💰

Read more: Gizmodo

🚧 NVIDIA Guardrails AI

Large Language Models (LLMs) are a double-edged sword. They allow us to answer complex life questions, write scathing emails in a “friendly tone,” create & debug code, and so much more. But they also pose a severe safety and security risk if not monitored and cleaned correctly.

NVIDIA decided to try and help reduce that risk.

Yesterday, they released an open-source tool, NeMo Guardrails, to ensure that generative AI applications are accurate, appropriate, and safe from potential security risks.

NeMo supports three broad categories of guardrails:

1. Topical Guardrail: one that ensures conversations “stay on topic” and don’t veer off-course

2. Safety Guardrail: one that ensures there is no “misinformation, toxic responses, or inappropriate content” during conversations (get this into Snapchat asap)

3. Security Guardrail: one that prevents LLMs from “executing malicious code or calls to an external application” that pose security risks.

This is big. After all the conversations we’ve had about AI safety and trustworthiness, these guardrails are a giant step in the right direction - especially given the recent scrutiny of generative AIs, such as Google's Bard and Microsoft Bing, for inaccuracies in their models.

According to NVIDIA, NeMo Guardrails is suitable for all LLMs, and any software developer can use it. The tool is available through the company’s AI Enterprise software platform and its AI Foundations service.

Try it out - it might just be the solution all you developers are looking for.

Read more: Engadget

🤗 HuggingFace announces chatbot

ChatGPT alternatives are growing by the day, and as DJ Khaled would say, here’s another one.

HuggingFace CEO Clem Delangue took to Twitter to announce HuggingChat, an open-source alternative to ChatGPT.

HC currently runs on OpenAssistant's latest LLaMA-based model and can handle basic tasks like writing code, generating emails, or writing songs. It’s an early prototype, so do not expect it to be perfect.

They are very open about their privacy details, which we like:

1. This ‘V0’ app version uses an anonymous cookie session. If you clear cookies, all the conversations will be lost.

2. You do not need an account to use the chatbot. Even if you are logged in, the app has no access to your HuggingFace account, and the conversations are only stored for the user, not for research or model training purposes.

3. In future versions, they are considering introducing a setting to allow users to share their chats with OpenAssistant to improve the training data and the model.

HuggingFace created this app to showcase that it’s possible to “build a ChatGPT alternative,” and it succeeded in doing so.

They say competition breeds innovation. If that’s true, we hope to see some major strides in conversational AI in the next few months. 🚀

Read more: TechCrunch

🤑 AI Fundraising News

CarbonChain Raises $10M - a UK-based company that provides an AI-driven carbon accounting platform.

Hakimo Raises $6M - a Menlo Park-based physical security technology firm.

🗞️ Byte size: AI article summaries

Disclaimer: AI is (partially) used to summarize these articles.

Republicans respond to Biden's re-election announcement with an AI-generated attack ad [The Verge] - The Presidential race is off to a fiery start. After Biden announced his reelection campaign, the RNC responded with an ad featuring AI-generated images portraying crises such as explosions in Taiwan after a Chinese invasion and military deployments on US streets if Biden is reelected in 2024. It raises concerns about the line between misinformation and regular campaigning and whether politicians should be exempt from fact-checking. (Read more)

Yelp rolls out AI-powered search updates and the ability to add videos to reviews [TechCrunch] - Yelp has incorporated AI-powered search via an improved algorithm and video integration with reviews. The enhanced AI-powered search will result in highlights featuring relevant information under each business listing. (Read more)

🐦️ Tweet of the day

🤯 😱

And that does it for today's issue.

As always, thanks for reading. Have a great day, and see you next time! ✌️

— Haroon: (definitely) Not A Robot and @haroonchoudery on Twitter