😮Corporate espionage overruns AI

ALSO: Another GPT level AI is released

Estimated Read Time: 4 minutes

Corporate espionage, whistleblowers, and new ChatGPT contenders! We’ve got it all today with headlines that preview the future struggle of the AI industry.

Today’s Horizon

• Chinese companies steal Google’s AI secrets

• Inflection-2.5 can spar with GPT4

• Stability tried to steal Midjourney’s data 

• Whistleblower raises the alarm at Microsoft

• AI is in more demand than tech jobs  

News (3 minutes)

😱Google’s AI secrets stolen in corporate espionage

What happened: Ex-Google engineer, Linwei Ding, was arrested by the FBI for stealing and trading secrets related to Google’s AI technology and secretly working with two Chinese firms. Classic corporate espionage.

Keep in mind all of this info is alleged. He has yet to be proven guilty.

The details: 

• Ding began transferring confidential information about Google’s supercomputing data centers to his Google Cloud account in May 2021 (at least use a different service).

• Ding worked on Google’s AI supercomputing systems and uploaded over 500 files over the course of two years.

• Ding was offered a CTO position at a Chinese startup and founded his own Chinese-based AI firm. All hidden from Google.

• Google has not commented on the potential impact of the stolen information.

Why it matters: The AI industry is in its land-grab phase. The arrest of a Google engineer stealing data for Chinese firms is an alarm bell for China's ambitions. China has been aggressively bolstering its tech sector, with a clear aim to close the gap with the US in artificial intelligence. Expect to see more espionage stories unfold in the coming years.

Further Details: 

• Ding is facing four counts of trade secret theft related to Google's AI technology.

• Ding faces a maximum penalty of 10 years in prison and a fine of up to $250,000 for each count, if he is convicted.

Pro Tip: Don’t store the top-secret data from the big mega corporation in the cloud service run by the big mega corporation.  

Stocks

📈AI Company Stocks Today

Has the Nvidia hype died down?

As of 03.08.2024 market close.

Headlines (2 minutes)

🚄SILVER BULLETS

AI is increasing in demand as tech jobs go down in demand. Tech jobs are 25% below pre-pandemic levels. Reports indicate demand has gone down at least 2.7% in the last three months. This combines with the recent series of high profile layoffs in the tech industry, some of which let go of thousands of employees.

AI in India will need government permission before launching thanks to new regulations. AI outputs will also need to be labeled with any potential biases or errors. Obviously the AI industry is critical of these regulations and argue they will stifle innovation. None of these regulations are a law yet, but show the future of legislation even if they do not pass.

We’re resurrecting the dead now. A Marilyn Monroe chatbot made by Soul Machines, an AI technology firm, can hold a conversation while using the long-dead actress’s voice and “realistic” facial expressions. And as always, it’s calling back to GPT 3.5 (of course). But this example really shows how one day people will “resurrect” long-dead loved ones with AI technology, making mimics of their dead husbands or wives, and even using their voices. In fact, I think this is inevitable (and creepy).

Future of AI (2 minutes)

😭Inflection-2.5 offers a shoulder to cry on, competes with GPT-4

What happened: Inflection just released Inflection-2.5, another AI model with almost parity to ChatGPT-4. It’s not quite there, but it’s actually good enough for most people. This is the second competitive AI model to release this week, the first being Anthropic’s Claude 3 Opus.  

The details:

• Inflection specializes in being more personal and emotionally intelligent than other AI models.

• The new update brings Inflection’s performance almost up to pace with GPT-4.

• Inflection is best used for more natural conversations that require a high EQ.  

• Inflection-2.5 is available to all Pi chatbot users, so if you use the assistant you can use Inflection’s newest model now.

  

  Source: Inflection’s Website

Why it matters: One of the big problems with AI models is their… emotional stupidity. Combine this with tons of dating and emotional support apps calling back to GPT-3.5, and you have AI’s telling people to kill themselves. Inflection’s idea of having an AI focused on high EQ, rather than pure IQ, is not only interesting but also sorely needed. 

Further details: 

• The AI offers improved performance in coding, mathematics, and offers real-time web search.

• Inflection raised over $1.3B in June 2023.

• Inflection has over 1 million daily user, so it’s gaining popularity.

• Conversations last an average of 33 minutes  

News (2 minutes)

🥷Midjourney bans competitor’s employees after corporate espionage

What happened: Midjourney just banned the employees of their competitor, Stability AI, after accusing Stability of image theft. Basically more corporate espionage.

The details: 

• Midjourney and Stability AI are fierce competitors in the AI image generation space.

• Midjourney had a 24-hour outage last weekend and blames Stability’s “bot attack”.

• Midjourney claims that Stability was using Midjourney’s images to train their AI (the cycle begins).

• Stability has denied any allegations (of course).

Why it matters: Tension continues to grow in the art community over AI generated art. Even more troubling is the thought of training AI art on other AI generated art. It’s like a snake eating its own tail. The accusation is ironic considering Midjourney itself is accused of image theft for training its AI on other people’s art. Time will tell if AI stagnates art or reinvigorates it.

Key-Concept (1 minute)

🛑Whistleblower raises the alarm for Copilot Designer

What happened: An engineer just alerted the FTC and U.S. Senate about Microsoft’s Copilot Designer, which is basically Bing’s Image Generator.

The details: 

• The whistleblower claims Copilot was producing sexual and violent images and was using copyrighted materials.

• Copilot Designer was introduced last year as Bing’s Image Generator.

• Jones claims Microsoft's legal team pressured him to delete a LinkedIn post on the issue.

• Microsoft has acknowledged his concerns.

Why this matters: AI is flawed. We saw that earlier last month with Gemini’s outrageous racial diversity among Vikings and Popes. Who trains the AI and what you train that AI on determine what the AI will ultimately create. Whistleblowers are the early warning system for problems coming our way. And this problem with AI images isn’t going to disappear anytime soon.

Productivity

🛠️AI Tools

• Replicate, a web interface for running various AI models. Essentially an API for running AIs.

• Beeyondai, an AI powered virtual assistant which allows you to train your own assistant.

• Mindsmith, an AI tool that helps you create and share content quickly. Specifically for creating lessons.

AI Generated Images

🥷Corporate Espionage!

Gotta stay in theme with today’s episode of AI wars.

Source: DALLE and Midjourney

We’re at your beck and call!

Got feedback? Rate today’s newsletter by clicking below! 

Got more personal feedback? Fill in the box when you rate us and give us your criticism and feedback.

Thank you for reading! 

❤️Share the (ARTIFICIAL) Love!

Got a friend who wants to appear as smart as you? An AI fanatic or someone shaking in their boots of getting replaced? Share us by clicking the link below and you’ll get our love ;)