🤦 They already hacked Microsoft’s “Recall”

ALSO: Former AI employees want the “Right to Warn”

Estimated Read Time: 10 minutes

We’ve got a doozy today. Massive AI news surged in as former OpenAI employees demand the right to warn the public about dangerous AI. And Microsoft’s Recall function is… already hacked. You can see everything that was one someone’s screen in just two minutes! What a mess. Buckle in and let’s go!

• 🤦 Microsoft’s “Recall” already got hacked.

• 🇮🇳 India just endured the world’s first AI election.

• 📢 OpenAI researchers want the “Right to Warn” about dangerous AI.

• 🦾 Musk diverts AI chips from Tesla to xAI and X/Twitter.

• ⛈️ Microsoft’s Aurora AI predicts the weather.

• 🎥 The “Netflix” of AI.

• 🕵️ Amazon’s AI Private Investigator.

• 😵 Microsoft is laying off more employees.

🤦 Microsoft’s “Recall” already got hacked

What happened: That was fast. Microsoft’s big AI feature, “Recall”, uses AI to watch everything happening on your screen by taking screenshots every five seconds. A pretty powerful feature except for one flaw: There’s already a tool which extracts all of the data Recall has collected. Bravo, Microsoft. 👏

The details: 

• The tool was developed by ethical hacker 🦸, Alex Hagenah, who wanted to prove a point; Recall’s screenshots are stored in an unencrypted database, making them vulnerable to attack.

• The tool is called “TotalRecall” (yeah, the movie) easily and automatically finds Recall’s database and extracts all of the data in plain text format.

• It takes two seconds to extract the data, including the screenshots of literally everything displayed on your screen. 👀

• The feature has been compared to spyware by, well… everyone who cares about privacy.

• Microsoft claims you can turn the feature off completely.

• Cybersecurity researcher Kevin Beaumont has built a site which uploads and searches extracted Recall information (he hasn’t released the site yet).

• Microsoft’s security holes have led to several US government data breaches.

Extra Spice 🌶️:

• Recently every major AI chatbot went offline at the same time. ChatGPT, Claude, Gemini (according to user reports) and Perplexity. Nobody knows why. It could be a security breach/DDOS attack, or a server side update. Hmmmm… 🤔

• HuggingFace recently reported unauthorized access (AKA a breach) in Spaces, their platform for sharing, hosting and developing AI models. They suspect private user information was accessed.

• If you are seriously interested in an AI-proof white-collar job, cybersecurity is looking very promising.

Why it matters: One of the biggest concerns for ALL of AI is cybersecurity. Even corporations (like Sony) and governments get hacked as we speak. AI is a powerful tool which uses private data to make decisions. What happens when these AI systems, which are integrated into daily life, are also hacked? Not to mention the obvious privacy concerns if this information is so easily available to Microsoft themselves. A screenshot of everything on my screen? Creepy.

A screenshot from “Leave the World Behind”, whose script was partially overseen by Barrack Obama. In the story the US undergoes a serious cyber attack, shutting down all major systems and creating an apocalyptic scenario. Then all of the self driving Teslas are hacked into crashing at high speed in order to block the roads.

What happens if a criminal hacks a civilian’s car and drives them off course to get kidnapped/robbed? If that sounds ludicrous, it shouldn’t. These are real possible consequences to having automated systems.

🇮🇳 India just endured the world’s first AI election… and it was pretty bad. Deepfakes galore. In the lead-up to India’s general election, Arvind Kejriwal, the chief minister of Delhi, was arrested for alleged corruption. He then used an AI voice clone to rally his followers from jail. And throughout the whole election, AI generated deepfakes ran amok. The impact was less than expected, but I think this is because AI tech is still early, and you can tell when something is fake for now. Politicians even used official AI voices in robocalls and voice messages to reach supporters.

🧠 Intel announced new AI chips to compete with Nvidia and AMD. The new Xeon 6 processor promises improved performance and power efficiency for data center workloads. I doubt this will hamper Nvidia though. The biggest difference with Intel is they also manufacture the chips themselves, so they have more control and possibly better pricing.

🚔 A man in Wisconsin was arrested for using AI to create child pornography. It was bound to happen sooner or later. Steven Anderegg allegedly used the AI image generator Stable Diffusion to produce over 13,000 explicit images of minors, none of which depicted real children. Additionally, he is accused of sharing a bondage-themed AI-generated image of children on Instagram and instructing a 15-year-old boy on how to use Stable Diffusion to create similar content (geez dude).

There’s been critique that the point of arresting these perps is to stop children from being abused and molested, and since the porn is AI generated he shouldn’t be arrested. But there’s an obvious problem with AI imagery… you can’t always tell if it’s real. Arresting someone for creating child porn, regardless if it’s AI generated, is a safer bet than trying to analyze if the porn is real. Especially as tech gets better and better and blurs the line with reality. Also it’s plainly disgusting and creepy. Nobody wants these images to exist anyway. I fully expect AI generated porn and creepy shit to increase over time.

📢 OpenAI researchers want the “Right to Warn” about dangerous AI

What happened: Several ex-employees of OpenAI and DeepMind have stepped forward to deliver a message demanding the right to warn the public about dangerous AI developments.

The details:

• The letter demands the right of AI insiders to report dangerous developments to the government and civil society since companies have financial incentives to keep hush-hush about AI’s risks.

• The petitioners openly reveal their names and mostly work for OpenAI, with some members of Google’s DeepMind as well.

• The letter claims that, without proper government regulation, it is up to insiders to hold these companies accountable.

Here’s what they demand:

• No NDAs/confidentiality agreements, as whistleblower laws wouldn’t protect insiders since AI is not regulated enough for dangerous activities to be illegal yet.

• There must be an anonymous reporting process.

• The company must be open to criticism.

• The company will not retaliate against insiders (current and former) for reporting their concerns to the public.

• Another OpenAI researcher recently left the company. And his reason for leaving? The drive for profits and “progress” over safety and benefitting humanity. He also signed the letter on righttowarn.

Why it matters: AI’s safety concerns are reaching a boiling point, especially with OpenAI. It seems every few days another “dangerous AI” report comes out from insiders. If this isn’t resolved soon, something is bound to blow.

If this many people are reporting about OpenAI, then something seriously dangerous is happening.

🦾 Musk diverts AI chips from Tesla to xAI and X/Twitter

What happened: Internal emails revealed Elon Musk just diverted $500M worth of AI chips from Tesla to xAI and X, which will delay Tesla’s own AI development.

The details: 

• Musk an order of 85,000 Nvidia chips by year-end and planned to spend $10 billion on Tesla’s AI development.

• Internal Nvidia emails revealed Musk redirected 12,000 AI chips from Tesla to X, delaying Tesla’s shipment of $500 million in GPUs by months.

• Critics argue he has a conflict of interest, relocating assets from one company to another (that’s a fair critique).

Why this matters: Tesla’s stock is down 29% as sales underperform. Perhaps Musk is simply diverting resources where they’re likely to be useful… and AI is where everyone hedges their bets. But that’s definitely a conflict of interest.

⛈️ Microsoft’s Aurora AI predicts the weather

What happened: Microsoft is introducing their new AI model, Aurora, promising to be the most advanced weather predictive AI in the world. That remains to be seen, but early impressions from tests look promising.

The details:

• Aurora is trained on over one million hours of weather simulations and supports 1.3 billion parameters.

• Aurora can forecast various atmospheric variables, including temperature, wind speed, air pollution, and greenhouse gas concentrations.

• Aurora uses a 3D Swin Transformer with 3D Perceiver-based encoders and decoders. Advanced tech for high definition forecasting.

• So far the model outperforms traditional numerical weather-prediction systems (what we’ve always used), offering 5,000x the computational speed-up over the Integrated Forecasting System (IFS).

• There’s actually a real massive disparity in traditional forecasting methods vs new AI models like Aurora.

• Bottom line: This AI model promises to predict weather better than anything else we’ve used in history.

Takeaway: A fantastic use for AI which could prevent a lot of tragedy. It’s also an obvious improvement. AI is perfect for predicting weather since weather prediction is based on raw data and correlation. Just one example of how AI will impact our lives, often without us even knowing it.

🎥 The “Netflix” of AI

What happened: Fable Studios has released the “Showrunner” app, allowing you to watch completely AI-generated TV show episodes in the style of South Park. Users can also generate an episode from scratch and customize the storyline to their liking. You can watch the first episode here. It looks like crap though.

The details:

• Everything is AI generated. From the storyline to the animations and voices. And… it shows (it’s not great right now).

• “Exit Valley” is their premiere show and features a South Park-esque art style.

• You can generate your own episode with prompts, describing the storylines and characters as you want.

• They’re targeting 22 episodes per season, and if a user’s episode is good enough they’ll include it in the cannon.

Takeaway: This might be the future of TV shows. Imagine telling a service what you wanted to watch, and the show is generated for you on the spot. Although that’s a far way off from being any good…

🕵️ Amazon’s AI Private Investigator

What happened: Amazon introduced Project P.I. (Private Investigator), an AI designed to investigate products to see if they are in perfect condition. It’s an interesting take on quality assurance.

The details:

• P.I. uses AI and computer vision to scan products for defects, such as damage or incorrect items, before they reach customers.

• Millions of items pass through imaging tunnels everyday where Project P.I. evaluates them. Defective items are isolated for further investigation (AKA a human).

• The system is currently active in several North American fulfillment centers, with plans to expand more this year.

Takeaway: If Amazon is doing this, then other companies will be sure to follow. Amazon is known for efficient supply lines, and it seems they found another way to cut costs using AI. Companies are eager to automate any process possible.

😵 Microsoft is laying off more employees

What happened: Microsoft is reportedly laying off hundreds of Azure Cloud employees despite the division’s growth and importance.

The details:

• They’ve confirmed around 1,000 layoffs affecting the Azure Cloud and Hololens mixed reality teams.

• This is despite Azure Cloud’s sharp growth due to AI tech, and the overall rise of Microsoft’s stock, income, and billions thrown around in investments.

• Microsoft already laid off 10,000 employees a year ago, including over one thousand from Activision/Blizzard, whom they recently acquired for $69.7B.

• Salesforce and Amazon have also announced big layoffs.

Takeaway: It’s sad that Big Tech goes through such huge growth and firing cycles. The tech industry has been going through thousands of layoffs over the past year, putting many people in a desperate position to get new jobs. Remember: Become irreplaceable within your company/job. You’ll need to do more than the bare minimum if you don’t want to get replaced by AI later on.

😱 Uncanny Valley

Source: Midjourney

Give us your feedback!

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 ;)