⚙️ EPIC calls on FTC to investigate OpenAI

Author

Ian Krietzberg
October 30, 2024

Good morning. Robert Downey Jr. — AKA Iron Man — intends “to sue all future executives" that attempt to create and use an AI-generated deepfake replica of him.

Even when he’s dead, he said, his “law firm will still be very active.”

— Ian Krietzberg, Editor-in-Chief, The Deep View

In today’s newsletter:

AI for Good: UCLA’s new medical vision model 

Source: UCLA Health

Researchers at the University of California, Los Angeles (UCLA) recently developed a new deep-learning model for the analysis of 3D medical images. 

The details: Unlike other medical imaging models, this model — nicknamed SLIViT, for SLice Integration by Vision Transformer — has a wide range of adaptability. It’s able to process 3D images from optical scans, CT scans, MRIs and ultrasound videos. 

  • In early studies, the research team found that the system is capable of accurately predicting disease risk factors from medical scans with only a moderately sized (and labeled) dataset. 

  • It does so by “leveraging prior ‘medical knowledge’ from the more accessible 2D domain.”

Why it matters: Researchers said that this could improve diagnosis efficiencies and timeliness, in a boon to patients and clinicians alike. The team plans to conduct several additional studies on the system. 

This is the Biggest Breakthrough in Next-Gen Tech Since iPhone—but you have less than 24 hours left to invest.

Fun fact: Did you know Disney's princess IP alone has generated $46.4B in revenue!? Wild.

Here’s another fun fact: a next-gen tech company called Elf Labs has won 100+ trademark battles for characters like Cinderella, Little Mermaid, and Snow White—yes, really—and has advanced proprietary tech to bring them to life right in your living room through next-gen AR (without clunky headsets). 

With their unprecedented compression and patented tech, they’re creating hyper-realistic, AI-powered 3D worlds with beloved, billion-dollar characters.

Now get this—Elf Labs has opened up an investment opportunity to the public. The team has already done $6B+ in licensing deals in their careers, and they're launching three new princess-based franchises in 2025. 

Time is running out, though: With less than 24 hours left to invest in Elf Labs, the opportunity to become a part of this game-changing movement ends today, October 30, at 11:59 pm PST.

OpenAI CFO says AI is no longer ‘experimental’

Source: OpenAI

OpenAI CFO Sarah Friar told Bloomberg this week that “AI is happening right now. It’s not experimental. It’s not just something people are playing around with.” 

The details: Friar said that OpenAI’s largest enterprise verticals are education and healthcare, though she added that financials — such as investment banks — are “probably third.” 

  • Saying that artificial general intelligence (AGI) is probably coming sooner than people expect — this is not a surprising line from her — Friar added that there’s a “lot of value in the product today.” 

  • She said that she watched a lawyer use OpenAI o1 to create a legal brief. The lawyer allegedly told Friar that he would have paid a paralegal between $1,000 and $2,000 per hour to do the same work. The ramifications of using a hallucinatory, biased system for legal work notwithstanding, the average hourly pay for paralegals is around $31 per hour, so not sure where the cost savings are here. 

Despite all this, OpenAI’s push into the enterprise doesn’t seem to be going very well. Friar said that 75% of its business comes from consumers — of OpenAI’s 250 million weekly active users, 5-6% have converted to paying users (at $20+ per month). 

It’s not clear exactly how active those users are (number of queries or session duration per week, for instance). 

A recent paper estimated that “between 0.5 and 3.5 percent of all work hours in the U.S. are currently being assisted by generative AI,” translating to an increase in labor productivity of between “0.125 and 0.875 percentage points.” 

Princeton computer science professor Arvind Narayanan called it a “glacial pace of adoption.”

OpenAI is on track to lose $5 billion in 2024, despite earning more than $3 billion in revenue. 

  • AI boom thrusts Europe between power-hungry data centers and environmental goals (CNBC).

  • Open-source AI must reveal its training data, per new OSI definition (The Verge).

  • GitHub partners with Anthropic and Google (TechCrunch).

  • US election betting ramps up in final stretch (Semafor).

  • AI automation arrives for salespeople (The Information).

If you want to get in front of an audience of 200,000+ developers, business leaders and tech enthusiasts, get in touch with us here.

  • Singlebase: A cloud platform for building AI apps and tools.

  • Happily: A tool to streamline the work environment.

Google earnings: AI is ‘paying off’ 

Source: Google

Reporting third-quarter earnings Tuesday night, Google parent Alphabet came in well above both earnings and revenue expectations, with CEO Sundar Pichai saying that the company’s investments in AI, long-term focus and commitment to innovation are “paying off.” 

Here’s how they did: 

  • Google reported revenue of $88.27 billion, a 15% year-over-year increase that came in above analyst expectations of $86.30 billion. 

  • Google reported earnings of $2.12 per share, above the $1.85 expected by analysts. 

Google Cloud revenue — a key metric for analysts — increased 35% to $11.4 billion, compared to expectations of $10.88 billion, an indication of ever-increasing AI demand. 

“The momentum across the company is extraordinary,” Pichai said in a statement. “In Search, our new AI features are expanding what people can search for and how they search for it. In Cloud, our AI solutions are helping drive deeper product adoption with existing customers, attract new customers and win larger deals.”

Shares of Google — up around 21% for the year — closed Tuesday up nearly 2% and rose more than 5% in after-hours trading as of Tuesday evening. The stock is valued at more than $2 trillion. 

This, all, despite the fact that the Department of Justice is considering breaking up the tech giant in light of a recent ruling that Google holds a monopoly in Search. 

Analysts respond: Deepwater’s Gene Munster called the results “triple good news” for Google: “Search is benefiting from generative AI, Google Cloud growth accelerated and margins were better.”

  • Regarding Search, this quarter marked the first full quarter of Google’s somewhat controversial AI Overviews; according to Munster, the generative AI search addition seems to be helping Search revenue. 

  • And, importantly, the 35% Cloud growth “is a positive read on the broader AI trade … this is evidence that hyperscalers’ cloud businesses are growing faster than expected and that growth is driven by compute for AI.”

Check out Rendora AI, the new tool we used to generate our recap video for last week's edition

Rendora AI is the world's first text-to-3D animated avatar platform. With Rendora, you can create your own avatar, input text, and generate a video in minutes — making it easier than ever to transform blocks of text into captivating visual experiences.

If you're interested in a free membership or if you're a newsletter owner who wants a free videotized preview, DM @rendora.ai on Instagram. Don’t miss out on the opportunity to elevate your content. Register at Rendora AI.

EPIC calls on FTC to investigate OpenAI

Source: OpenAI

The Electronic Privacy Information Center (EPIC) on Tuesday filed a complaint with the Federal Trade Commission (FTC) requesting an investigation of OpenAI for several violations of the FTC Act. 

The organization said that OpenAI has failed to meet established public policy standards for the responsible development and deployment of its generative systems by “offering products with unsafe security, privacy and business practices, perpetuating unfair and deceptive practices in their product development and release and causing significant consumer harm.”

EPIC said that there is no evidence — or that OpenAI is choosing not to provide that evidence — that its products were developed safely and responsibly, in line with existing law. 

“The generative AI boom promised that it could create all kinds of benefits to society. Thus far, it has created a mass data-stealing regime and countless privacy, human rights, data security and accuracy problems,” Calli Schroeder, lead of the AI and Human Rights Project at EPIC, said in a statement. “The hypothetical benefits do not outweigh the actual harms and OpenAI must be accountable for what their product is actually doing — not what they hope it will do.”

OpenAI did not respond to a request for comment. 

Before we get into the complaint, Section 5 of the FTC Act prohibits “unfair or deceptive acts or practices in or affecting commerce.”

The allegations: The cornerstone of EPIC’s complaint here is that OpenAI’s products “were developed via unprecedented web-scraping and the collection, utilization and dissemination of stolen data from millions of consumers.” 

  • Acknowledging the few licensing deals that OpenAI has inked with publishers, EPIC claimed that most of OpenAI’s training data comes from massive web crawls performed  “without informed consent or a single notification to consumers.”

  • The organization said that this data includes personally identifiable information (PII) concerning millions of people; of particular note, they wrote, is the likely quantity of PII “released by people other than the subject of that personal data — like family members posting photos” online. 

EPIC added that there is a high risk that this personally identifiable information is at risk due to security concerns associated with the company and its models. 

The organization further argued that the hallucination and bias inherent to Large Language Models (LLMs) — including OpenAI’s systems — pose enormous risks of discriminatory outcomes. 

Tying it all together: EPIC said that OpenAI is in violation of Section 5 of the FTC Act, arguing that its “indiscriminate web scraping, widespread data retention, development of biased algorithms and failure to adopt appropriate data security and accuracy testing measures are unfair because they cause, or are likely to cause, substantial injury to consumers.”

EPIC wants the FTC to investigate these points, then require the company to stop its unfair data scraping practices, facilitate model evaluation and testing and destroy any models that were made using “illegally collected” consumer data, among other things.

At the core of this is an overwhelming lack of transparency coming from OpenAI and most of the other major labs. Though we suspect — and sometimes, this information is leaked and published by certain outlets — that OpenAI uses certain types of information, we don’t know the details of its training data, which means that no one can effectively understand the details of the biases inherent to its models, or the scope of PII infringement on display here. 

It will be very interesting to watch how this progresses if the FTC does choose to investigate OpenAI based on these claims. A result could be the forceful unveiling of these opaque details to the public; not sure if that would change anything, but change starts with knowledge, and that would certainly grant us knowledge regarding, at least, training data, which would absolutely change the game. 

Still, I would be very surprised if model destruction results from this.

Which image is real?

Login or Subscribe to participate in polls.

🤔 Your thought process:

Selected Image 1 (Left):

  • “Imperfect carving on the pumpkins had to be real. Plus the light shining down on the AI pumpkins seemed too staged.”

Selected Image 1 (Left):

  • “The faces of the Jack O' Lanterns in Image 2 are too basic. Like the AI just Googled Jack O'Lantern and said ‘Yeah this will fool those dumb humans!"‘ NOT TODAY AI!!!”

Not today. Well done.

💭 A poll before you go

Thanks for reading today’s edition of The Deep View!

We’ll see you in the next one.

Here’s your experience with medical AI transcription tools:

For a third of you, your doctor has not yet attempted to use AI transcription tools.

But a quarter of you said their doctor has used such tech; another 20% said their doctor wanted to, but they declined to give permission.

Are you buying Google stock after that report?

Login or Subscribe to participate in polls.