Jack Hidary, the CEO and founder of Sandbox AQ, is no stranger to using quantum ideas. He even authored a book on it, entitled Quantum Computing: An Applied Approach. His research group has also been spun out as a separate corporation after six years of operating under Alphabet, Inc. (Google’s parent firm).

While quantum computers are extremely beneficial, they are still a few years away. Sandbox AQ aims to tackle contemporary problems using artificial intelligence and quantum physics without having to wait for quantum computers.

“What we discovered is that AI and quantum technology can be used now, that it can be run today, that it can work today, for businesses and major corporations,” Hidary added.

Cryptography that is quantum-resistant
Indeed, one of Sandbox AQ’s first major endeavors is in the field of cybersecurity, which is a problem that must be solved before quantum computers are powerful enough to break today’s codes.

The information must sometimes be kept secret for decades by governments, defense contractors, financial institutions, and healthcare facilities. That means that secure communication sent today using current encryption standards like 2048-bit RSA (Rivest–Shamir–Adleman) could be intercepted and saved for future quantum decryption to gain access to the secret data.

Many organizations, including the United States National Institute of Science and Technology (NIST), have been working on post-quantum cryptography, also known as quantum-resistant cryptography, for years. They’re working on new cryptographic system standards that will be secure against future quantum computers while also integrating with existing protocols and networks.

This six-year global, multi-stakeholder process, according to Hidary, is nearing a “positive finish.” One of the issues that contributed to Sandbox AQ’s decision to split from Alphabet was the implementation of these standards.

Sandbox AQ believes that AI will be critical in the development of post-quantum cryptography.

“An AI model decides in real-time which protocols and parameters to utilize to encrypt and decrypt at that particular time,” Hidary explained. Their current commercial solution aims to increase security for billions of smartphones, laptops, servers, and Internet of Things devices around the world. “The good news is that we can accomplish it in software,” Hidary continued.

Data centers, point-of-sale devices, microcontrollers, and Internet of Things (IoT) devices will all require these security methods. In addition, the business has established partnerships with Vodafone and Softbank to develop virtual private networks that use post-quantum cryptography (VPNs).

“If you’re creating chips,” Hidary says, “you’ll need to enable these post-RSA encryption-decryption protocols.” “We’re in the midst of a generational shift.”

Furthermore, Sandbox AQ is employing AI to assist security audits in what they refer to as the “finding phase” in order to detect system flaws.

If you have a large enterprise network with many subnets, endpoints, devices, firewalls, routers, and so on, you’ll need to undertake a lot of investigation to figure out where the susceptible protocols are being used.”

Fault Tolerance is Required for Quantum Computing
Sandbox AQ’s overall goal is to keep a watch on quantum computing developments. Hidary, on the other hand, stated that they do not plan to deploy quantum processing units (QPUs) until they can scale to higher quantum bit (qubit) counts and become fault-tolerant by adding error correction.

He does see a “really interesting future” in which CPUs, GPUs, and QPUs collaborate in the cloud to perform what they do best.

Quantum communication and sensing are leading the way.
While the impact of quantum processors may be years away, Sandbox AQ is investigating quantum phenomena in the realms of sensing and communications. “Birds and some species of whales can navigate thousands of kilometers extremely precisely using the Earth’s magnetic field,” Hidary said, describing quantum magnetometry as a future complement to GPS navigation. If we can harness this potential, it could give a solid backup in situations where GPS is unavailable.

Sandbox AQ is also interested in quantum coherent communication. Hidary presented a new communications network that is being created as a parallel Internet but is based on quantum notions such as superposition and entanglement.

According to Hidary, it is being built now all over the planet, even though it will take many years to complete.

What is the best way for me to break into the AI and Quantum industries?
Hidary stressed the importance of Sandbox AQ’s support for training the future generation of AI and quantum engineers, scientists, and mathematicians during our chat. Partnerships with institutions throughout the world and a residence program for Ph.D. students and post-docs are among Sandbox AQ’s initiatives.

“We provide generous pay and a lot of training that is complementary to what they would receive in academics.” They also learn that world-class science and research can be conducted in a commercial setting.”

Consistent Advancement Towards a Quantum World
While quantum processors receive the majority of the attention, Hidary envisions a world in which quantum processors, quantum sensors, and other quantum devices all communicate with one another through a quantum coherent network. Sandbox AQ will be intriguing to watch in the approaching quantum world, with its high-profile coming-out celebration and support from a wide range of sectors.

But what is deep learning precisely, and how does it work?

Deep learning is a subtype of machine learning that makes use of neural networks to learn and predict. Deep learning has demonstrated outstanding performance in a variety of tasks, including text, time series, and computer vision. Deep learning’s success is largely due to the availability of massive amounts of data and computing power. Deep learning, on the other hand, is considered superior to any of the traditional machine learning algorithms for a variety of reasons.

Neural networks and functions in deep learning
A neural network is a network of neurons that is interconnected and each neuron is a limited function approximator. Neural networks are seen as universal function approximators in this sense. A function, as you may recall from high school algebra, is a mapping from input space to output space. A basic sin(x) function maps from angular space to real number space (-180o to 180o or 0o to 360o) (-1 to 1).

Let’s have a look at why neural networks are regarded as universal function approximators. Each neuron learns a limited function: f(.) = g(WX), where W is the input vector, X is the weight vector, and g(.) is a non-linear transformation. WX can be represented as a line in high-dimensional space (hyperplane), with g(.) being any non-linear differentiable function such as sigmoid, tanh, ReLU, and so on (commonly used in the deep learning community).

Finding the best weight vector W is the essence of learning in neural networks. We have two weights in y = mx+c, for example, m and c. Now, we identify the best value of m & c based on the distribution of points in 2D space that meets certain criteria: the gap between predicted y and actual points is minimal for all data points.

The effect of layers
We stack numerous such neurons in a “layer” where each neuron receives the identical set of inputs but learns distinct weights W. As a result, each layer has a set of learned functions termed hidden layer values: [f1, f2,…, fn]. In the next layer, these values are combined once more: h(f1, f2,…, fn) and so on. Each layer is made up of functions from the previous layer (for example, h(f(g(x))). It has been demonstrated that any non-linear complex function may be learned using this composition.

A neural network with several hidden layers (typically more than two hidden layers) is known as deep learning. Deep learning, on the other hand, is essentially a complicated synthesis of functions from layer to layer, with the goal of identifying the function that specifies a mapping from input to output. If the input is an image of a lion and the output is an image classification indicating that the image belongs to the lion’s class, deep learning is learning a function that maps image vectors to classes. Similarly, the input is the word sequence, and the output is whether the input sentence is positive, neutral, or negative. As a result, deep learning is the process of learning a map from input text to output classes, which might be neutral, positive, or negative.

Interpolation with deep learning
Humans process images of the world by hierarchically understanding them bit by bit, from low-level elements like edges and contours to high-level features like objects and sceneries, according to a biological interpretation. In neural networks, function composition is similar, with each function composition learning complicated characteristics about an image. The most common neural network design for images is the Convolutional Neural Network (CNN), which learns those features in a hierarchical form and then classifies them into distinct classes using a fully connected neural network.

We try to fit a curve through interpolation that somewhat represents a function defining those data points, utilizing high school math once more. Given a set of 2D data points, we try to fit a curve through interpolation that somewhat represents a function defining those data points. The more complex the function we fit (for example, in interpolation, determined by polynomial degree), the better it fits the data; however, it generalizes less for a new data point. This is where deep learning runs into problems with what is known as an overfitting problem: fitting to data as closely as possible while sacrificing generalization. Almost all deep learning architectures had to deal with this critical feature in order to train a generic function that can perform equally well on unknown input.

“Deep Learning is not as stunning as you think because it is just interpolation coming from glorified curve fitting,” wrote Yann LeCun (developer of the convolutional neural network and ACM Turing award winner) on his Twitter handle (based on a paper). However, there is no such thing as interpolation in high dimensions. Everything at high dimensions is extrapolation.” As a result, deep learning does nothing but interpolation or, in some situations, extrapolation as part of function learning. That concludes our discussion.

The educational aspect
So, how do we learn such a difficult function? Well, it all relies on the situation at hand, and the neural network architecture is determined by that. When it comes to picture classification, CNN is the tool of choice. We utilize RNN or transformers if we want to make time-dependent predictions or messages, and we use reinforcement learning if we have a dynamic environment (like driving a car). Aside from that, learning entails dealing with a variety of challenges:

1. Regularization is used to ensure that the model learns a general function rather than only fitting to train data.
2. The loss function is chosen based on the task at hand; roughly speaking, the loss function is an error function between what we want (actual value) and what we currently have (current prediction).
3. Gradient descent is the procedure for convergent to an optimal function; determining the learning rate is difficult because when we are far from optimal, we want to move faster to optimal, and when we are close to optimal, we want to move slower to ensure we converge to optimal and global minima.
4. The vanishing gradient problem necessitates a large number of hidden layers; architectural adjustments such as skip connections and an appropriate non-linear activation function aid in solving it.

Challenges in computing
Now that we know that deep learning is nothing more than a learning complicated function, it presents a new set of computational challenges:

a) A significant amount of data is required to learn a complex function.
b) We need fast computation environments to process enormous amounts of data.
c) Such settings necessitate infrastructure that can support them.

To compute millions or billions of weights, parallel processing with CPUs is insufficient (also called parameters of DL). Learning weights for neural networks necessitates vector (or tensor) multiplications. GPUs can be particularly handy here because they can do parallel vector multiplications quickly. We sometimes require 1 GPU, and sometimes numerous GPUs, depending on the deep learning architecture, data size, and task at hand. A data scientist must make this decision based on published literature or by assessing performance on one GPU.

A deep learning network may learn any mapping from one vector space to another vector space given the right neural network architecture (number of layers, number of neurons, non-linear function, etc.) and enough data. Deep learning is a strong tool for any machine learning endeavor because of this.

One of the most obvious advantages of global connection is the capacity to converse and transact with anyone, wherever in the globe. Global connectedness nowadays is primarily based on the internet. According to some estimates, there were 4.95 billion active internet users worldwide in January 2022, accounting for 62.5 percent of the global population.

In the world of global communication, satellite internet is gaining popularity. Satellite internet is expected to replace traditional cable internet in the near future because of its ease of use, worldwide coverage, and independence from terrestrial infrastructure. ‘Satellite internet refers to the method of connecting to the internet via communication satellites, most commonly geostationary satellites’ (now low-Earth satellites also). Unlike cable internet, which uses intercontinental submarine cables and fiber optics to connect users, satellite internet uses three basic interconnected components to form a network: network operation centers (NOCs), communication satellites, and a dish antenna connected to a modem at the user’s end.

Data is transmitted via cables from Internet Service Providers (ISPs) to NOCs in the aforementioned network. NOCs, which serve as ground stations, provide data through radio waves to satellites, which then send the same data to users all over the world. The data is received from the satellite via dish antennas at the user’s end, and it is translated by the attached modem into a format that can be used by PCs and other devices. To complete the data cycle, the same channel is used to send (upload) data from the user end to the NOCs, and then to the Internet.

Since the entrance of private enterprises in commercial space-related operations, satellite internet’s dominance in the global internet industry has been rapidly increasing. Because of its accessibility in remote locations, satellite internet saw a considerable increase in use during the Covid-19 outbreak.

According to Allied Market Research, the global satellite internet market was valued at roughly $2.93 billion in 2020, and based on current trends, it is expected to reach $18.59 billion by 2030, with a CAGR of 20.4 percent. So far in 2022, HughesNet, Viasat, and Starlink are the top three satellite internet providers. HughesNet and Viasat’s services are mostly reliant on geostationary satellites orbiting 22,236 miles from Earth, which is a considerable distance. Furthermore, many businesses solely provide services in the United States.

SpaceX, a well-known spaceflight business, has launched Starlink, a satellite internet enterprise that will supply services around the world via a giant constellation of thousands of Low-Earth Orbit (LEO) satellites. The US Federal Communications Commission (FCC) has already given Starlink authorization to launch 12,000 satellites, and it is seeking permission to launch 30,000 more in the near future. Various national space agencies, as well as environmentalists, have expressed concern over space pollution as a result of this. Even NASA has written to the FCC, expressing worries about the possibility of satellite collisions in space due to the Starlink satellites.

Several news stories about Starlink being prevented from delivering its services in Pakistan circulated in Pakistani media in late January. “… satellite broadband operator Starlink has never sought for nor secured any license from PTA to operate and supply internet services in Pakistan,” the Pakistan Telecommunication Authority (PTA) said on January 19, 2022. In Pakistan, there is currently no national legal framework in place to regulate space-related activities and services. The Ministry of IT and Telecom has been in discussions with Starlink officials about the policy and operation of satellite internet access in Pakistan, but no significant progress has been made.

Despite technological and environmental problems, satellite internet is quickly gaining popularity among people all over the world. In the near future, it will change the way homes, small and medium companies, and community internet services work. It will also contribute to closing the digital divide between the developed and developing worlds by allowing both to interact and conduct business. The internet services supplied by satellite constellations, however, must be strictly regulated at both the global and national levels.

International regulatory organizations must ensure that private-satellite launches do not add to space junk/debris or interfere with other satellites’ important space functions. Similarly, the Pakistani government must provide regulatory standards for the operations, registration, licensing, and taxes of space-related technology services in the country in order for them to be widely used.

According to Pavel Zavalny, “friendly” countries could be permitted to pay in crypto or their own currencies.

Russian President Vladimir Putin indicated earlier this week that he wants “unfriendly” countries to buy Russian gas in roubles.

The action is thought to be intended to strengthen the Russian ruble, which has lost nearly 20% of its value this year.

Following the invasion of Ukraine, sanctions implemented by the United Kingdom, the United States, and the European Union have put pressure on Russia’s rouble and risen the cost of living.

Russia, on the other hand, remains the world’s largest natural gas exporter and second-largest oil supplier.

Mr. Zavalny, the chairman of Russia’s State Duma’s energy committee, said on Thursday that the country is looking into alternate payment methods for energy exports.

China and Turkey, he said, were among the “friendly” nations that were “not involved in the sanctions pressure.”

“We’ve been recommending to China for a long time that we convert to national currency settlements for roubles and yuan,” Mr. Zavalny added. “It’ll be lira and roubles with Turkey.”

“You can even exchange bitcoins,” Mr. Zavalny continued.

‘More danger’

Despite the hazards, analysts believe Russia would profit from accepting the popular cryptocurrency.

“Unprecedented sanctions are having an immediate impact on Russia,” said David Broadstock, a senior research fellow at the Energy Studies Institute in Singapore. “There is a need to strengthen the economy, and Bitcoin is considered as a high-growth asset in many aspects.”

However, he pointed out that Bitcoin’s value has fluctuated by as much as 30% this year. In comparison, the dollar has been trading within 5% of the euro.

“Clearly, accepting Bitcoin creates significantly more risk in the transaction of natural gas as compared to other traditional currencies,” Mr. Broadstock added.

“Moreover, China is one of Russia’s most important ‘friendly’ economic partners, and Bitcoin is prohibited in China,” he continued. “This certainly limits the use of Bitcoin for payment.”

There are fears that Russian oligarchs would use virtual currency to circumvent sanctions.

This has prompted the Ukrainian government, as well as US and European politicians, to request that all Russian users be barred from crypto-currency sites.

However, many businesses have ruled this out.

“Now that their currency has collapsed, some regular Russians are using cryptocurrencies as a lifeline,” said Brian Armstrong, CEO of cryptocurrency business Coinbase.

“Many of them are likely to reject what their country is doing, and a ban would harm them as well,” he said.

Mr Putin’s remarks on forcing “unfriendly” countries pay in roubles sent the currency to a three-week high on Wednesday.

However, many existing gas contracts are in euros, and it is uncertain if Russia will be able to change them. For 40% of its gas, the EU is reliant on Russia.

According to WABetaifo, the app is working on implementing end-to-end encryption in its call logs. WhatsApp plans to release the latest update for Android version 2.22.8.7 via the Google Play Beta Program.

What is end-to-end encryption, and how does it work?

End-to-end encryption means that neither WhatsApp nor its parent company, Meta, can listen in on your conversations.

WhatsApp’s installation of the sign is an attempt to remind its users that WhatsApp is a safe place to communicate with their friends and family.

A similar footer might be added to the bottom of the chats list and status updates. However, there is currently no evidence of this.

The end-to-end encryption symbol can already be found across the WhatsApp application. This tool also aids in the encryption of your private discussions. The communication can only be seen by the sender and receiver.

The official announcement reads as follows:

Cumulative Update Build 22581.100 is now available for download (KB5013283). This release contains no new features and is intended to test our Dev Channel build servicing pipeline.

As a result, all of the bug fixes and enhancements that were announced with the previously released build 22581 to the Dev Channel and Beta Channel have been carried over. As we go towards the next phase of Windows 11 in preparation for version 22H2, the build offers a lot of new features.

While there are no new features in this build, servicing test builds like this have been known to break several functionalities in the past. For example, x64 emulation on ARM PCs was broken in Windows 10 Build 21292.1010.

You can read the original announcement here. https://blogs.windows.com/windows-insider/2022/03/23/announcing-windows-11-insider-preview-build-22581/

Okta reported that in the “worst case,” 366 of its clients were affected and that their “data may have been accessed or acted upon” – the company’s stock dropped 9% as a result of the announcement.

It claims to have over 15,000 customers, ranging from large corporations like FedEx to smaller businesses like Thanet District Council in Kent.

The breach was carried out by the cyber-gang Lapsus$.

According to Ekram Ahmed of cyber-security firm Checkpoint, the ransomware gang “is a South American threat actor that has recently been linked to cyber-attacks on certain high-profile targets.”

“The cyber-gang is infamous for extortion, threatening the publication of sensitive information if its victims do not comply with their demands,” he said.

The gang has claimed to have broken into a number of high-profile companies, including Microsoft, in the past.

Microsoft stated in a blog post that Lapsus$ had only gotten restricted access after compromising a single account, but that no customer code or data had been compromised.

Concerns grew.
The attack in January, according to Okta, was carried out by a third-party contractor known as a “sub-processor,” and “the situation was probed and contained.”

“Aside from the activities observed in January, there is no evidence of continued malicious activity,” it stated.

However, as the public’s worry grew, Okta issued a series of updated blog postings that provided more information.

Over a five-day period in mid-January, hackers gained access to the computer of a customer-support engineer working for the sub-processor, according to Chief Security Officer David Bradbury.

He described the attack as “analogous to walking away from your computer at a coffee shop, where a stranger has sat down at your machine and is utilizing the mouse and keyboard,” he said.

However, the engineer’s computer did not grant “god-like access,” the hackers were limited in their actions, and Okta was not compromised and remained fully operational.

“Our clients are not required to take any corrective activities,” Mr. Bradbury continued.

The engineer’s employer, Sykes, a Sitel Group company, said it was “certain there is no longer a security concern.”

However, it would “continue to examine and assess any security vulnerabilities to both our infrastructure and the brands we support around the world” in partnership with external cyber-security specialists.

In online remarks, Lapsus$ claimed that it had not stolen “any databases from Okta” and that it was solely focused on its clients.

Although no difficulties have been detected by Okta’s clients, Mr. Ahmed advises “great caution and cyber-safety practices.”

“In the following days, the full breadth of the cyber-resources gang’s should be revealed,” he continued.

A variety of applications
Cloudflare, one of Okta’s clients, stated in a blog post that it did not believe it had been hacked.

“There is no indication that our environment has been hacked or compromised,” FedEx told Reuters.

Thanet, which uses Okta to make it easier for employees to manage and sign in to different apps, told BBC News that the attack “has not affected the council’s data’s security,” but that it “will continue to monitor the situation.”

The National Cyber Security Centre in the United Kingdom said it has “not observed any evidence of impact in the United Kingdom.”

According to a regulator, the internet giant participated in “false, misleading, or deceptive conduct” by knowingly displaying adverts for fraudulent cryptocurrency.

Financial and other penalties could be imposed on the US corporation.

Meta has yet to respond but has previously stated that it is dedicated to preventing scammers from using its services.

The ads in question, according to the Australian Competition and Consumer Commission (ACCC), exploited Facebook’s algorithms to target vulnerable consumers and featured phony comments from Australian celebrities.

Former New South Wales Premier Mike Baird, popular TV broadcaster David Koch, and rich entrepreneur Dick Smith were among the names used without permission.

The ACCC chairman Rod Sims said in a statement on Friday that “the essence of our argument is that Meta is liable for these adverts that it posts on its platform.”

Meta allegedly did this intentionally and failed to prevent the scams despite celebrity concerns, according to the court action filed in Australia’s Federal Court.

“We know of one consumer who lost almost A$650,000 (£360,000; $480,000) as a result of one of these frauds being falsely presented as an investment opportunity on Facebook. This is a disgraceful situation “Mr. Sims explained.

Last month, Australian billionaire Andrew Forrest filed a criminal complaint against Meta for using his image in fake ads.

While Dr. Forrest accuses Google of violating anti-money laundering regulations, the ACCC’s complaint involves suspected violations of consumer law or another regulatory act.

In 2021, Meta, which also owns Instagram and Whatsapp, generated $115 billion in global advertising income.

Why? Because Google employed a novel method of determining the relevancy of the results returned by a query. It was successful in performing an automated peer evaluation of websites. The more pages that linked to a site, the more relevant it was thought to be, and thus it was given a higher rating. The algorithm that performed this, known as PageRank, was the foundation upon which Google’s dominance of internet search was created.

Google swept all before it because its ranking system appeared to be objective: it just counted links and ranked them accordingly. Of course, it can be manipulated, and a cottage industry of search engine optimizers has sprung up to ensure that their clients’ pages rank highly in Google searches. However, Google consumers may rest assured that the corporation was not favoring certain results over others. There was no advertising involved.

The company’s founders were adamant that favoring sponsors’ pages would taint the results’ accuracy. “We expect that advertising-funded search engines will be inherently biassed towards advertisers and away from the needs of consumers,” they wrote in 1998. “We believe the issue of advertising causes enough mixed incentives that having a competitive search engine that is transparent and in the academic realm is crucial.”

That’s correct. After burning through investors’ money for two years, the founders realized in 2000 that high-minded neutrality doesn’t pay the bills, so they evolved into surveillance capitalists, monitoring their users to collect information about them that marketers would find interesting. Google’s sales surged by approximately 4,000 percent between then and the company’s initial public offering in 2004.

For a long time, the incursion of advertising concerns didn’t seem to worry users much, despite the fact that it irritated industry competitors and authorities, particularly in the EU (to which Google has paid billions of euros in fines). Despite this, it has maintained its dominance as the most used search engine in many parts of the world for, well, forever. That shows the corporation is doing something well, if only because true rival search engines are easily available, unlike, say, Facebook.

All of this makes a blog post by Dmitri Brereton, a software programmer at a San Francisco firm, all the more intriguing. “If you’ve tried to look for a recipe or product review recently, I don’t need to tell you that Google search results have gone to shit,” Brereton wrote under the heading Google Search Is Dying. You’ve probably seen that the first few non-ad results are SEO [search engine optimization] optimized sites with affiliate links and advertisements.”

“Especially when it comes to true information,” he adds, Google still provides good results in many other categories. You may believe that Google results are beneficial to you and have no idea what I’m talking about. What you don’t realize is that you’ve been preventing yourself from searching for the majority of the things you’d like to search for. You already have a sneaking suspicion that Google isn’t going to give you a nice outcome.”

My hunch is that this is the online equivalent of a storm in a teacup because I don’t have a dog in this battle (I use Google very infrequently and DuckDuckGo most of the time). When I use Google, it’s usually for real information, so my experience may differ from that of the Reddit and Hacker News crowds. It’s possible, as the New Yorker posits, that Google’s results are a reflection of how good the SEO camp is at gaming PageRank.

Three additional plausible reasons for displeasure with Google, according to the CEO of DuckDuckGo (who, of course, has a dog in the fight). One is people’s dislike to being followed. Another source of irritation is Google’s preference for its own items in purchase-related search results. And what about the third? Simply boredom: we’ve been in AG for so long that people are yearning for a change. If that’s what’s really bothering them, they should keep in mind that help is only a click away.