Happy Monday — and welcome back to Big Brain.

We’re like a double shot of espresso. Only you don’t have to worry about a serious caffeine crash (or digestive problems) after.

Becoming an AI-expert is a reported symptom though…

🔦 Today's AI spotlights:

• Using AI for crime

• Meta is making AI moves

🎁 REMINDER: We’ve just released 4 AI tool kits for: PMs, Designers, Engineers, Marketers.

To get a free copy, just share our newsletter with 1 person (details at bottom).

🔦 USING AI FOR CRIME

We’ve seen the ways that AI can be used for good.

But how about the ways it can be used for bad?

This past week we saw 2 scary examples where people used AI to commit crimes 👇

Example 1: Breaking into a bank account

Joseph Cox wrote an article about how he used an AI-generated voice tool to break into his bank account.

In a YouTube video, Joseph can be seen accessing the Lloyd Bank voice ID sign-in program.

The program asks for various voice prompts. Instead of saying them himself, Joseph plays audio clips that had been generated by training on his voice.

Example 2: Fake luxury yacht party

Later in the week, a Chinese yacht part organizer was busted for using AI.

The scammer promised “VIP maid service” to attendees.

However, when customers scanned the promotional QR code, they noticed something was wrong.

The 43 portraits of the “maids” had weird looking hands, giving away that they had been created with AI…

The takeaway from both these stories is that we should expect a rise in AI-fueled crimes.

Its not clear that current security measures are good enough to detect them. And the problem will only get worse as AI programs improve.

🔦 META IS MAKING AI MOVES

Microsoft, Google, Amazon and Apple have all been making big moves in AI.

But one big tech company has been eerily quiet on the AI front. That company is… Meta.

For most of the past year, Zuck and co have been all in on the idea of “the Metaverse”. (They’ve poured $36 billion into it since 2019)

However, it appears the company is shifting its attention to the hottest area in tech.

Last week the company announced the release of a new large language model called “LLaMA”.

LLaMA is short for “Large Language Model Meta AI”.

The model will be available to researchers, governments, and academic entities.

Meta claims that the model requires far less computing than previous offerings and has trained on 20 languages. It also has 65-billion paramaters.

It will also be interesting to see how Meta leverages its proprietary data set to gain an advantage in AI.

With billions of users, Meta can train on content that competitors do not have access to.

📰 OTHER HEADLINES & STORIES

• How One Guy’s AI Tracked the Chinese Spy Balloon Across the US

• Marc Andreessen and Steven Sinofsky on AI sentience, ethics, love, jobs

• Qualcomm demos fastest local AI image generation with Stable Diffusion on mobile

• South Africa’s Envisionit Deep AI gets $1.65M to expand access to medical imaging

🛠️ AWESOME AI TOOLS

• tldrbot: AI assistant that summarizes Discord chats with one command

• SpiritMe: Become a digital avatar in 5 minutes

• Audioshake: Take any song and break it into its stems

• Replai.so: Easy-to-use AI replies generator for Twitter and LinkedIn

🖼️ AI ART PIECE

Dream sink — by Kate Norlund

🤖 CHATGPT BING PROMPT OF THE DAY

Some users are reporting that Bing is not producing essays for them anymore? If you’re on the beta, please reply with your results!