One in four AI learners from non-tech backgrounds in India: Report

One in four AI learners from non-tech backgrounds in India: Report



One in every four Indians learning artificial intelligence (AI) now comes from a non-technical background and one in five belongs to Tier-II cities, a new study has found, indicating that AI is rapidly evolving from a specialist technology skill into a mainstream workforce capability and reshaping careers across industries.

 


According to the India AI Workforce Report 2026, released by AI-native technology company Scaler on Thursday, nearly 25 per cent of AI learners come from non-engineering fields, while around 50 per cent of AI-enabled career outcomes are emerging outside traditional engineering roles.

 


The study, based on data from 11,444 professionals across India, claimed that AI is becoming a workforce-wide capability, with professionals across industries using it to improve productivity, accelerate career progression and create new opportunities.

 
 


“AI is no longer a tool only for software engineers and technology teams. While much of the global discourse around AI focuses on the threat of job losses, the analysis suggests it is creating opportunities rather than destroying jobs. The technology is becoming a catalyst for career growth,” said Abhimanyu Saxena, co-founder of Scaler.

 


The report revealed that AI upskilling has led to an average salary increase of 147 per cent across experience levels. Early-career professionals reported the highest percentage gains, with salary growth averaging 155 per cent. Professionals with six to nine years of experience saw average compensation increase by 140 per cent, while those with nine to 12 years of experience reported average salaries touching ₹38 lakh annually after AI upskilling.

 


It was also found to be reshaping India’s leadership pipeline, as 27 per cent of those with AI knowledge attained leadership roles, including engineering, data science and data engineering leadership positions. Learners moving into leadership positions were also found to command some of the highest salaries, with vice-presidents, CXOs and engineering leaders earning average annual compensation of around ₹33 lakh after AI-focused learning.

 


Software engineering remains the most common AI career outcome, accounting for nearly 35 per cent of placements and career transitions. Consulting has emerged as another major beneficiary of AI adoption, with the share of consulting outcomes nearly doubling from 3.1 per cent among learners to 5.65 per cent among post-upskilling professional outcomes.

 


Beyond career growth, the study also found a significant democratisation of AI talent across the country. While Bengaluru continues to lead India’s AI ecosystem with 19 per cent of learners, followed by Pune, Hyderabad, Mumbai and Chennai, the AI talent pipeline is expanding significantly into smaller cities.

 


“Nearly one in five AI learners now comes from Tier-II cities such as Lucknow, Jaipur, Patna, Indore, Coimbatore, Chengalpattu and Nagpur. Access to AI education and remote work opportunities is helping bridge the historical gap between metropolitan and non-metropolitan talent pools,” the report stated.

 


The study also highlighted the growing role of women in India’s AI economy. Women are steadily leveraging AI skills to break into technology and business roles that were previously less accessible. Women professionals who transitioned into AI-enabled careers reported an average salary increase of 145 per cent.

 


Women working as quality assurance engineers reported salary jumps of as much as 574 per cent following AI upskilling. Significant gains were also recorded among women in engineering leadership, machine learning engineering, backend engineering and data science roles. They were also found to be helping expand AI’s influence across non-engineering functions.

 


The report argued that India is uniquely positioned to emerge as a global AI talent powerhouse due to its large technology workforce, thriving digital ecosystem and young population eager to embrace new skills. AI learning is creating a workforce that is more diverse, more inclusive and capable of driving innovation across industries.

 


“What excites us most about the study is that the real transformation is taking root in Tier-II cities, among women professionals and across functions far beyond engineering. AI is creating new pathways to opportunity, accelerating career growth and enabling professionals to command stronger compensation outcomes,” Saxena added.

 



Source link

GenAI may add -17 bn to real estate GVA over next seven years: Report

GenAI may add $14-17 bn to real estate GVA over next seven years: Report



Generative artificial intelligence (GenAI) could add $14-17 billion to the real estate sector’s gross value added (GVA) over the next seven years, equivalent to a 3-4 per cent uplift in real estate value, according to a joint report by EY-Parthenon and Credai.

 


Developers could see a 30-50 per cent improvement in sales velocity and around 30 per cent faster product launches with the help of GenAI.

 


The report also finds that GenAI can cut deal evaluation time by about 50 per cent, reduce land-closure turnaround time by 30-35 per cent, and enable 2.5x more deals to be evaluated through automated feasibility modelling, seller assessment, and internal rate of return (IRR)/return on investment (ROI) scenario generation.

 
 


These possibilities stem from AI-driven customer intelligence, automated design workflows, and predictive project monitoring, signalling a shift from scale-driven operations to a more intelligence-led approach to real estate, the report said.

 


Further, early adopters could experience strong operational benefits, including a 20-50 per cent improvement in workforce productivity, 20-50 per cent lower customer acquisition costs, and decision cycles that may compress from “months to weeks or days”.

 


The report noted that such advancements have the potential to reshape how developers assess feasibility, plan projects, manage construction, and engage with customers.

 


Chaitanya Seth, partner, real estate practice, EY-Parthenon India, said, “We see GenAI-led transformation unlocking 2-3x enterprise value within the short to medium term, by compressing land-to-launch cycles by 20-30 per cent, driving 30 per cent-plus sales acceleration, and delivering a 5-20 per cent step change in efficiency across cost and timelines.”

 


Seth emphasised that this is not about incremental digitisation; rather, it is about rewiring the operating model, redefining customer experience, strengthening brand advocacy, and building brands that scale faster and sell smarter.

 


Shekhar Patel, president, Credai, said, “The next phase of growth in Indian real estate will be driven not only by scale, but increasingly by intelligence, speed, and the ability to make better decisions across the project lifecycle. Going forward, the focus should be on harnessing these capabilities to build a smarter, more efficient, and resilient real estate sector that can deliver greater value to homebuyers and support India’s urban growth ambitions.”

 



Source link

Tim Cook's price hike warning puts Apple's feeder lineup at risk: Analysts

Tim Cook's price hike warning puts Apple's feeder lineup at risk: Analysts


Apple plans to raise prices on its products to offset rising memory and storage chip costs, said company CEO Tim Cook in an interview with the Wall Street Journal published on Wednesday. Cook did not say when prices would go up, by how much, or which products would be affected first — but the direction is now on the record.

 


“Unfortunately, price increases are unavoidable,” Cook told WSJ. “We’re doing our best to mitigate the huge increases that are being passed to us, and we’ve been trying to shield our customers from the increases, but the situation has become unsustainable.”

 
 


The interview comes after months of visible strain on Apple’s product lineup — from Mac configurations being quietly delisted to delivery timelines stretching into months. Cook’s remarks are the clearest signal yet that the company is moving from managing the situation internally to passing costs on to customers.


The memory problem


The root cause is something that has been building for over a year. AI infrastructure is consuming memory at a scale the industry wasn’t built for. Data centres running large language models require high-bandwidth memory (HBM) — a specialised type of DRAM — and manufacturers like Samsung and SK Hynix have been steadily shifting production capacity toward it because the margins are substantially better. The result is a supply squeeze on general-purpose DRAM and NAND flash, the kind that goes into smartphones, tablets, and laptops.

 


“There’s less supply at a time when consumers want devices and the memory guys are passing along huge price increases,” Cook told the Journal. “We definitely need memory pricing and supply to return to reasonable levels for consumer products. That’s the bottom line.”


Cook indicated that Apple is particularly exposed on the DRAM side, singling out the LPDDR market as a specific concern. He also noted that Apple is prepared to use its cash reserves to help secure memory supply, without going into specifics. “We’re willing to use our balance sheet to help be a part of the solution,” he said. “Obviously, more capacity is needed.” He was careful to clarify, however, that Apple has no plans to build its own memory or storage factories.

 


Cook will hand over the reins to John Ternus in September. This interview may be one of the more consequential things he says before he does.


This was coming


The signals were there months before Cook put it on the record.

 


In January, ZDNET Korea reported that Samsung and SK Hynix had raised LPDDR prices for Apple by close to double compared to the previous quarter — Samsung reportedly proposed an increase of over 80 per cent, while SK Hynix proposed roughly 100 per cent.

 


Apple analyst Ming-Chi Kuo flagged the implications at the time. In a post on X, he noted that iPhone memory pricing had moved to quarterly negotiations from the previous six-month cycle, and that another hike was already visible for the second quarter of 2026. Kuo also wrote that Apple’s plan for the upcoming iPhone was to absorb as much cost as possible and keep starting prices flat — at least for marketing purposes — while accepting margin compression.

 


By February, after Apple’s Q1 FY2026 earnings call, Kuo summarised the situation: Apple had publicly identified memory as a “cost pressure” rather than a “supply constraint.” That distinction mattered to investors tracking semiconductor stocks. On the earnings call itself, Tim Cook acknowledged that memory had a minimal impact on Q1 gross margins but would be more of a factor in Q2 — a signal that the numbers were already starting to move.


Already visible on the Mac lineup


Price hikes are still ahead, but supply constraints are already here, and buyers have been living with them for months.

 


As reported by Business Standard in May, several Mac Studio, Mac mini, and MacBook configurations were already facing delays or had been quietly delisted from Apple’s online store in India. What started with high-memory Mac Studio variants in April — where 512GB unified memory options disappeared and 256GB variants showed four-to-five month delivery windows — had, by May, spread to the MacBook line. MacBook Air models with 32GB unified memory were facing weeks of delivery delays. The MacBook Pro with 128GB unified memory was showing roughly a month of lead time.


The pattern was consistent: higher-memory variants vanish first, then mid-tier configurations start stretching out, and by the time you notice, the options have quietly narrowed.


Apple’s structural advantage, and its limits


Apple is not a normal customer for memory suppliers. It ships approximately 247 million iPhones a year, according to a Business of Apps report. iPhone DRAM and NAND consumption accounts for something in the range of 20 to 25 percent of global smartphone memory demand, as per Kuo. That scale gives Apple leverage that only a few consumer device companies can match.

 


Counterpoint Research, in a note published on June 16, made Apple’s advantage explicit. In week 20 of 2026, global smartphone sales fell 8 per cent year-on-year — the ninth consecutive week of decline. Apple was up 10 per cent over that same period, while Chinese OEMs like Xiaomi, OPPO, and Vivo were down between 10 per cent and 19 per cent.

 


Counterpoint’s Associate Director, Sujeong Lim, noted that brands with stable supply chains and high visibility into memory were able to maintain more consistent pricing and promotional strategies, calling Apple’s position in that regard “advantageous.”

 


Faisal Kawoosa, Founder and Chief Analyst at Techarc, echoed a similar stance “Apple is not only tier 1 customer for all the supply chain but a marquee one as well for everyone supplying them components including memory. At the same time it doesn’t operate in segments like sub Rs 30k in smartphones where BOM structure is completely disrupted. So from both negotiations powers and cost structure it’s well cushioned as a brand.”

 


Navkendar Singh, Associate Vice President at IDC India, put a finer point on what Apple’s leverage actually means in practice right now. Quantifying the buffer is difficult, he said, but the more important advantage is not price — it is supply assurance. “Even if somebody is willing to give the price, supply is assured to these brands more than the pricing.” Apple and Samsung, by virtue of their sheer volume and supply chain pull, are in a different conversation with memory manufacturers than everyone else, he added.

 


But scale has limits when the structural imbalance is severe enough.

 


Kawoosa said that Apple has always been tight on maintaining profitability, and as “increasing component costs eat up its profit margins, it will spill over some price increase to the customers.”

 


Kawoosa also connected the Mac delisting pattern to the broader cost environment: “Yes, due to memory costs, brands including Apple are having lesser variants of RAM-storage configurations.”


What gets hit first


Apple’s premium positioning gives it a buffer that mid-range Android brands don’t have, but that buffer does not apply uniformly across its own lineup.

 


Kawoosa flagged Apple’s so-called feeder portfolio as the most exposed. “For India and other emerging markets, I do see some concern around its ‘feeder portfolio’ — essentially the E series of smartphones, Neo for MacBooks and Air for Tablets — where we might get to see early impact.”

 


He added that he would not be surprised to see Apple introduce a Neo lineup for iPads, widening the entry-level tier in a way that gives it pricing flexibility without visibly raising prices on flagship models.

 


Singh added that Apple accessories such as Apple Watch and AirPods will likely remain insulated from the impact of component price inflation. He said that these devices “still run on legacy silicon, and the volumes are far lower” compared to the likes of iPhones, Macs and iPads.

 


“While you might see it somewhere, I think it will be more pronounced in these categories,” he said, referring to iPhone, Mac, and iPad.


Impact on possible foldable and September launches


Apple is reportedly on track to launch its first foldable iPhone in September alongside the iPhone 18 Pro and Pro Max. That is a significant new category, and it arrives in the middle of a component cost squeeze.

 


Singh does not expect the foldable to be shelved, but he does not rule out a minor slip in timeline. “Foldable, I’m not too sure that it will have an impact. Might be a delay of one or two months if at all. Otherwise, we’re looking at a foldable launch this year only.” He added that if it does not land in September alongside the Pro lineup, it will still happen by October or November at the latest.

 


As for the next generation iPhones, Singh said that higher storage tiers like 1TB will be available in smaller quantities, with supply weighted more toward 256GB and 512GB variants.

 


On pricing, Singh said Cook’s interview should be read partly as Apple setting the stage for price increases on the September launches. He added that buyers should expect roughly $100 to $150 upside on every SKU.

 


It should be noted that Bloomberg has reported that Apple is shaking up its release schedule. The company typically unveils all models in the new iPhone series at the same time in September. But this year, the September lineup will reportedly be limited to the higher-end models, such as the Pro, Pro Max, and the anticipated foldable. The report stated that the new standard iPhone model, along with the refreshed Air model, could launch in spring 2027.


What this means for India


While Apple’s premium offerings grab the most headlines, its feeder portfolio, as Kawoosa describes it, pulls first-time Apple buyers and young professionals into the ecosystem. If memory cost inflation makes those entry points more expensive, or thins out the configuration options available at existing prices, the calculus changes for a significant segment of buyers.

 


Singh’s read on India demand was measured. IDC’s forecast for the overall Indian smartphone market this year is a 13 per cent decline, but he said the bulk of that drop is coming from Android. Apple, he said, will not degrow in any meaningful sense. “Even if it degrows, it’s a plus minus few percentage points. It will hold because Apple customers are more price immune than other brands.”

 


Kawoosa was direct about what Apple’s exposure means at a market level: “Apple shrinking in demand will affect the value of the market more than volume.” Even a modest pullback in iPhone sales in India hits the premium segment hard because iPhones dominate the high average selling price end of the market.

 


This is even clearer from Counterpoint’s data. The market intelligence firm reported that in India, Apple had 28 per cent smartphone market wholesale value share in CY 2025 while having roughly 9 per cent market share in terms of shipments.

 

For buyers who have been sitting on the fence, that context cuts both ways. Apple’s pricing has held its ground far, but with cost increases now confirmed in direction if not in detail, the current generation of devices may represent the last relatively stable entry point for some time. If you have been planning to buy, now is probably the better side of the wait. 


Kawoosa put it plainly. “It is anyways the best time to buy any gadget from any brand right now because prices are only going up. Even if there are festive season discounts later, they will at best bring prices back to current levels. The traditional logic of waiting for festive offers no longer holds. People used to wait because they would get good deals. This time, prices are not in any control.”

 



Source link

Is multi-factor authentication enough? Kali365 breach fuels security debate

Is multi-factor authentication enough? Kali365 breach fuels security debate



For years, multi-factor authentication (MFA) has been one of the most widely recommended safeguards against account compromise. The logic was simple: even if attackers obtained a user’s password, they would still need access to a second verification factor to gain entry.

 


However, a phishing kit known as Kali365 is drawing attention because it targets something beyond passwords and authentication codes.

 

The US Federal Bureau of Investigation (FBI) has warned that Kali365, a phishing-as-a-service (PhaaS) platform, is being used to compromise Microsoft 365 accounts by capturing authentication tokens after users have already completed MFA verification. Rather than stealing credentials directly, the toolkit hijacks authenticated sessions, allowing attackers to access services such as Outlook, Teams and OneDrive as legitimate users.

 
 


The technique reflects a broader shift in cyberattacks, with threat actors increasingly targeting active sessions instead of login credentials.

 


The risks extend beyond a single account. Since Microsoft 365 often serves as a gateway to corporate email, files and business applications, a compromised session can provide access to sensitive data and internal communications.

 


Although the FBI first warned about Kali365 in May, the phishing kit has attracted renewed attention in recent weeks as researchers revealed more details about its operations and concerns grew around the increasing use of token-theft and session-hijacking techniques.


What is Kali365


In a recent public service announcement, the FBI’s Internet Crime Complaint Center (IC3) described Kali365 as an “emerging phishing-as-a-service (PhaaS) platform” that first appeared in April 2026.

 


According to the agency, the toolkit is distributed primarily through Telegram and is designed to help attackers access Microsoft 365 accounts by stealing authentication tokens, thereby bypassing MFA protections.

 


Phishing-as-a-service refers to a criminal business model in which developers provide ready-made phishing tools and infrastructure to other cybercriminals for a fee.

 


The FBI said Kali365 offers features such as AI-generated phishing lures, automated campaign templates, real-time tracking dashboards and OAuth token-capture capabilities. It effectively packages sophisticated phishing tools into a subscription-based service.

 


Cybersecurity publication BleepingComputer, citing research from Arctic Wolf, reported that Kali365 operates much like a business. The platform has developers who maintain the service, resellers who market it and affiliates who launch phishing campaigns.

 


Researchers said the toolkit supports multiple attack methods, including techniques capable of capturing session cookies and authentication tokens even after a user has completed MFA verification.

 


Authentication tokens are digital credentials issued after a user successfully signs in. They allow users to remain logged in without repeatedly entering passwords or completing MFA checks. If attackers obtain these tokens, they can access an account as though they were the legitimate user, even without knowing the password.

 


According to security researcher Graham Cluley, writing on Bitdefender’s Hot for Security blog, access to Kali365 is reportedly available through a subscription model priced at around $250 a month or $2,000 a year.

 


Researchers also reported that hundreds of attacks linked to the toolkit were observed across North America and Europe within weeks of its emergence.

 


How the scam works

 


What makes Kali365 different from conventional phishing campaigns is that it does not focus on stealing passwords. Instead, it tricks users into authorising access to their accounts and then captures the authentication tokens that Microsoft issues after a successful login.

 


According to the FBI, the attack generally follows four stages:

 


Lure: The attack begins with a phishing email designed to resemble a legitimate notification from Microsoft or another trusted cloud service. The email asks the user to complete a sign-in process and provides a unique device code. Unlike many phishing scams, victims are not directed to a fake website. Instead, they are instructed to visit a genuine Microsoft verification page, making the request appear legitimate.

 


Authorisation: Once on the Microsoft page, the user enters the device code and signs in with their account. If MFA is enabled, they may also complete the additional verification step. At this stage, the victim believes they are simply logging into a service, but they are actually granting the attacker’s device permission to access the account.

 


Token theft: After the login is approved, Microsoft issues authentication tokens that prove the user has successfully signed in. These tokens allow services such as Outlook, Teams and OneDrive to recognise the user without repeatedly requesting a password. Kali365 is designed to capture these tokens, giving attackers the same level of access as the legitimate user.

 


Persistence: With the stolen tokens, attackers can continue accessing Microsoft 365 services even though they never learned the user’s password. Because the tokens indicate that MFA has already been completed, attackers can often bypass additional authentication prompts and retain access until the tokens expire or are revoked.

 


How widespread is this technique and is Kali365 unique

 


Kali365 is more like a broader trend rather than a one-off threat. According to reporting that cited researchers from cybersecurity firm Proofpoint, multiple device-code phishing kits with nearly identical tactics were observed within a span of just 10 days. Researchers noted that many of these campaigns appeared highly automated and likely generated with the help of AI, suggesting that adoption of the technique is growing rapidly among cybercriminal groups.

 


The same report noted that device-code phishing attacks were already being used to compromise Microsoft 365 accounts before Kali365 emerged. In December, researchers documented cases involving both state-backed threat actors and financially motivated cybercriminals using similar methods to gain access to accounts.

 


Further evidence of the trend came from researchers at Huntress and Flare.io, who earlier this year linked a separate wave of attacks to another device-code phishing platform known as “Evil Tokens.” The findings suggest that Kali365 is not an isolated tool but part of a growing ecosystem of phishing services designed to steal authenticated sessions rather than passwords.

 


Is MFA losing effectiveness

 


Multi-factor authentication (MFA) remains one of the most effective defences against account takeovers because it can stop attackers who have obtained a user’s password from gaining access. In that sense, MFA is still doing the job it was designed to do.

 


The challenge highlighted by Kali365 is different. Instead of trying to crack or bypass MFA, attackers wait for the legitimate user to complete the authentication process themselves. Once the user has successfully signed in and verified their identity, the attackers steal the authentication tokens generated during that session. Those tokens can then be used to access the account without needing the password or another MFA prompt.

 


Security experts therefore argue that the issue is not that MFA is broken, but that cybercriminals are increasingly targeting parts of the authentication process that occur after MFA has already been completed. This broader trend is reflected in industry research.

 


According to cybersecurity firm SentinelOne, which cited findings from the 2025 Verizon Data Breach Investigations Report (DBIR), MFA-fatigue attacks accounted for 14 per cent of analysed security incidents involving MFA bypass. In those attacks, users are bombarded with authentication requests until they eventually approve one.

 

Device-code phishing, the technique used by Kali365, works differently. Rather than overwhelming users with prompts, it abuses a legitimate Microsoft authentication workflow to obtain access that the user unknowingly authorises. 

 


The shift from passwords to 2-factor authentication to passkeys

 


The way people secure their online accounts has changed over time as cyberattacks have become more sophisticated. For years, passwords were the first line of defence. But passwords alone proved insufficient because they could be guessed, stolen in data breaches, reused across multiple websites or captured through phishing scams.

 


To strengthen account security, companies introduced two-factor authentication (2FA), also known as multi-factor authentication (MFA). This added an extra verification step, such as a one-time code, an authenticator app notification or a biometric check. Even if a password were compromised, attackers would still need a second factor to gain access.

 


However, cybercriminals have adapted. Instead of focusing solely on passwords, many now target the login process itself through phishing campaigns, MFA fatigue attacks and token theft. Tools such as Kali365 are part of this shift, allowing attackers to take advantage of authenticated sessions rather than stealing credentials directly.

 


As a result, technology companies are increasingly promoting passkeys as the next step in account security. Passkeys allow users to sign in using a fingerprint, face scan or device PIN, without relying on traditional passwords. Because they are linked to a specific device and website, they are considered more resistant to phishing attacks.

 

The growing attention around Kali365 highlights why this transition is underway. The challenge is no longer just preventing password theft but ensuring that attackers cannot misuse the trust established after a user has successfully logged in. 

 


What organisations can do now

 


The emergence of Kali365 is prompting organisations to look beyond traditional account-security measures. While multi-factor authentication remains an important safeguard, the attack demonstrates how cybercriminals are increasingly targeting authenticated sessions and access tokens rather than passwords themselves. As a result, security teams may need to strengthen protections around the entire login process, not just the initial sign-in.

 


In its advisory, the FBI outlined several measures to limit the attack techniques used by the phishing kit and reduce the chances of unauthorised access to Microsoft 365 accounts. These include:


  • Restrict device code authentication: According to the FBI, organisations should consider blocking or limiting device code authentication wherever possible. This is the login method that Kali365 exploits to trick users into granting access to their accounts.


  • Review existing usage before making changes: Before restricting device code authentication, companies should identify which applications, devices or workflows rely on it. This helps avoid disrupting legitimate services, such as conference-room systems and shared workplace devices.


  • Limit authentication-transfer features: The FBI also recommends restricting features that allow users to transfer authentication between devices. These workflows can create additional opportunities for attackers to exploit legitimate login processes.


  • Maintain emergency-access accounts: Organisations should ensure that emergency or break-glass accounts remain available and are excluded from broad restrictions. This can help administrators regain access if normal authentication systems become unavailable.


  • Strengthen phishing defences: The FBI further points organisations to guidance from the US Cybersecurity and Infrastructure Security Agency (CISA), which recommends employee awareness training, phishing detection measures and stronger identity-security controls.



Source link

AI startup Midjourney pivots to health with 'full-body ultrasound machine'

AI startup Midjourney pivots to health with 'full-body ultrasound machine'



By Chris Welch 
AI startup Midjourney Inc. announced its first hardware project at an event in San Francisco, outlining an unexpected move into the personal health and medical industries. 


Chief Executive Officer David Holz revealed what he described as a “full-body ultrasound machine” called the Midjourney Scanner. “No such device has ever been built until now,” he claimed, touting the new technology as superior to MRI (magnetic resonance imaging) in numerous ways.

 


“Our goal is to build a fleet of 50,000 of these scanners,” Holz said, declining to specify how much the product will cost. “We’re not even using any AI in this yet — just really cool hardware and software.”

 
 


The machines, which require users to be partially submerged in water, will debut in “Midjourney Spa” locations, with the first to come in a 25,000-square feet space in San Francisco that will include hot tubs, saunas, cold plunges and a gym, among other amenities. “We’ve signed a lease and we already have the designs,” Holz said.

 


The device is one of eight projects that Midjourney is currently focused on, split between four hardware and four software initiatives. Holz said that the company aims to ship at least two of those hardware efforts in the near term.

 


Addressing potential hurdles that could come with meeting FDA requirements, Holz said “one of the goals is to do all things that are easy,” and layer on approval after approval “until this thing can do thousands of types of diagnoses,” he said. “Over a 10-year period, these things are not just imaging devices: They’re probably therapeutic as well.”

 


Before Wednesday’s sudden business shift, Midjourney was mostly known as a generative AI subscription service that lets customers create images and video clips. Prices range from $10 to $120 per month.

 


The startup has faced legal challenges from content owners including Warner Bros. Discovery Inc. and Walt Disney Co. over alleged theft of intellectual property, with the entertainment companies claiming that their well-known characters have appeared in Midjourney’s generated images and videos.



Source link

What is BGP hijacking? The protocol at the centre of Telegram's India clash

What is BGP hijacking? The protocol at the centre of Telegram's India clash



Reliance Jio has denied allegations made by Telegram founder Pavel Durov that the telecom operator was using a technique known as Border Gateway Protocol (BGP) hijacking to disrupt access to the messaging platform for users outside India. 


The denial came a day after Durov levelled the accusation on social media platform X, amid Telegram’s escalating dispute with Indian authorities over the government’s temporary restriction on the platform ahead of the NEET-UG 2026 re-examination. 

The Centre ordered a temporary block on Telegram until June 22, arguing that the platform was being used by organised cheating networks and to spread misleading information related to the examination. Telegram moved the Delhi High Court on June 17, calling the restriction disproportionate and arguing that it affected more than 150 million Indian users.

 


What Pavel Durov accused Reliance of


In his post, Durov alleged that Reliance was sabotaging access to Telegram for millions of users outside India, including in the United Arab Emirates (UAE), through what he described as a rogue technique called BGP hijacking. 


He further claimed that the issue appeared intentional because Meta owns roughly a 10 per cent stake in Reliance Jio Platforms, suggesting the telecom operator may have acted to give WhatsApp an advantage while Telegram remained restricted. 


Durov urged network operators to reject unauthorised route announcements that he alleged were originating from Reliance. 


Reliance Jio later issued a public rebuttal stating that it had “not been involved in any such incident” and that it continues to operate its network in accordance with global internet-routing best practices.


What is BGP and why does the internet need it?


Most people think of the internet as a single giant network. In reality, it consists of tens of thousands of smaller networks operated by telecom companies, cloud providers, governments, universities and technology firms. 


Whenever someone opens Telegram, YouTube or Gmail, their request typically travels through several of these networks before reaching its destination. The system that helps these networks find one another is called the Border Gateway Protocol (BGP). 


Think of the internet as a giant courier network. Every telecom operator, cloud provider and internet company manages its own 
delivery network. Telegram’s servers may be connected to one network, while your internet connection may come from another.


 
Suppose you want to send a parcel from Delhi to a small town in Kerala. The courier handling the parcel may not serve that destination directly. Instead, it passes the parcel to another courier company, which may pass it on again until it reaches the final address. 


For that system to work, each courier company must know which company can deliver to which locations. BGP performs a similar role on the internet. It allows networks to continuously exchange information about which internet addresses they can reach and how traffic should be forwarded. 


In effect, BGP functions as a global directory that helps internet traffic find its destination across thousands of interconnected networks. 


Without BGP, networks would not know where to send data once it left their own systems, making global internet connectivity virtually impossible.


Why is BGP vulnerable to hijacking?


The key issue is that traditional BGP does not have a built-in mechanism to automatically verify every route announcement it receives. 


If one network announces that it can deliver traffic to a particular destination, other networks may accept that claim and begin forwarding traffic accordingly. 


In simple terms, BGP often operates on trust first and verification later. 


This means that if a network accidentally publishes incorrect routing information, or deliberately announces routes that do not belong to it, traffic can be redirected through the wrong path. This is commonly known as a BGP hijack. 


Researchers and internet-governance organisations have warned about this weakness for years. According to the Internet Society-backed MANRS initiative, routing incidents such as route leaks and route hijacks continue to occur globally despite ongoing efforts to improve security.


Why was BGP designed this way?


BGP was developed in the late 1980s when the internet was far smaller and less complex than it is today. 


At the time, the number of connected networks was limited and participants generally knew and trusted one another. As a result, the primary goal was to ensure different networks could communicate efficiently. 


Security threats such as state-backed cyber operations, large-scale internet fraud, cryptocurrency theft and global social media platforms either did not exist or were not major concerns. 


Because of that, BGP was built around a trust-based model, with networks largely expected to provide accurate routing information. 


That design helped the internet grow rapidly but also created vulnerabilities that are still being addressed decades later.


Is this a one-off incident?


No. Routing incidents involving BGP have occurred for years across the global internet. While many result from accidental configuration errors rather than deliberate attacks, they can still disrupt services and redirect traffic through unintended networks. 


India has witnessed several such incidents. 


In 2015, incorrect route announcements involving Bharti Airtel briefly affected traffic destined for Google services. In 2020, researchers at Oracle Internet Intelligence reported route leaks involving Vodafone Idea. 


These incidents highlight a broader reality: because internet routing relies on thousands of independently operated networks exchanging information, mistakes can sometimes spread far beyond the network where they originated.



Source link

YouTube
Instagram
WhatsApp