ManageEngine Insights

According to economists from the World Economic Forum, AI-related debt pressures are a worrying macroeconomic trend in 2026. Worrying, indeed. I believe an AI bubble burst is absolutely in the cards, although the burst will likely occur after the Anthropic and OpenAI IPOs.

The AI bubble has been building for some time now. The “magnificent seven” tech stocks (Alphabet, Amazon, Apple, Nvidia, Meta, Microsoft, and Tesla) currently make up 33% of the S&P 500, and AI-related investment accounted for over 90% of U.S. GDP growth in the first two quarters of last year. The spending on AI infrastructure has shown no signs of stopping, with analysts at J.P. Morgan Chase predicting spending of $5 trillion more over the next four years. Four heavyweight hyperscalers—Alphabet, Amazon, Meta, and Microsoft—intend to spend $670 billion on AI infrastructure in 2026 alone.

The math problem underlying the AI boom

As German economist and mathematician Joachim Klement points out in his recent piece for the Financial Times, the spending on AI infrastructure dwarfs what was happening at the height of the dot-com crisis. Last year, U.S. businesses invested $1.5 trillion in IT software and hardware—an incredibly high number, especially when compared to the level of spending during the height of the dot-com bubble, which totaled $466 billion ($829 billion when adjusted for inflation).

Klement, a managing director at Panmure Liberum, believes the math behind the AI boom is impossible and will lead to a severe correction. He writes, “The US economy is growing solely because of the tech boom. I calculate that over the past four quarters, 93 percent of U.S. GDP growth was explained by tech investments. Even at the peak of the TMT (technology, media, and telecom) bubble, it barely reached 60 percent.”

According to Klement, if tech investments were to decline by 4-6%, the U.S. economy would enter a recession very quickly. And Klement doesn’t foresee a world in which the hyperscalers don’t have a negative return on investment over the next five years (he makes an exception for Amazon).

Speaking directly to the over-hyped IPOs of Anthropic and OpenAI, Klement writes, “The IPO of these AI companies is probably nothing more than a major transfer of investment risk from the current owners to retail investors, pension funds and others who are willing to buy the hype.”

Interestingly, Klement is well-known for picking the winning teams of the 2014, 2018, and 2022 men’s World Cup finals. However, for what it’s worth, he picked the Netherlands to win this year, so perhaps he’s not Nostradamus after all. Nevertheless, the fact remains: we’re in an AI bubble and it will likely burst.

Big tech is taking on a ton of debt

According to Moody’s Analytics, tech companies issued a whopping $108.7 billion in corporate bonds during the last quarter in 2025, and this trend has continued through the first half of 2026.

Mark Zandi, a chief economist at Moody’s, says, “It’s a lot of debt, and a lot of it all of a sudden.” When these big tech companies fund unproven ventures with debt, “it does put the broader financial system at risk. If the financial system is at risk, then the broader economy is.”

Speaking about hyperscalers like Google, Microsoft, and Meta, venture capitalist Paul Kedrosky says, “If these companies are so profitable, why are they using debt? It gives you a sense of the scale of what’s going on.”

Circular financing, cross-holdings, and questionable financial engineering

Writing in TIME Magazine, Ganesh Sitaraman, a law professor at Vanderbilt University, and Asad Ramzanali, the director of AI and tech policy at the Vanderbilt Policy Accelerator, recently sounded alarm bells about the questionable financial engineering practices that are so prevalent among today’s major AI players.

Sitaraman and Ramzanali write, “We’re seeing a rise in specific forms of financial engineering—circular financing, “off books” special purpose vehicles, huge private credit loans, and signiﬁcant volumes of credit default swaps and asset-backed securities—which obscure a full understanding of the systemic risks.”

The circular equity financing within so many AI infrastructure deals is particularly concerning. And we’re seeing a lot of this. Company 1 invests in Company 2; then Company 2 uses those funds to buy from Company 1. Think about all the cloud companies and chipmakers investing in Anthropic and OpenAI right now.

As Sitarama and Ramzanali note in their March 2026 publication “After the AI Crash”, vendor-based equity investments at this scale is a new phenomenon, which they describe as a “new form of financial engineering.” They warn, “All of the largest AI, cloud, and chip companies own parts of each other, which means a small or unexpected problem at one company can cascade quickly to all of them.”

We are also seeing cloud-for-credit exchanges and mark-to-market accounting rules being employed. Cloud-for-credit exchanges occur when hyperscalers like Microsoft and Google invest in AI companies like OpenAI and Anthropic and then record OpenAI and Anthropic’s returning cloud spend on their books as revenue.

Mark-to-market accounting allows the hyperscalers to book unrealized gains on their equity stakes as net income. According to Sitaraman and Ramanali, “Amazon’s stake in Anthropic alone added $16.8 billion to its Q1 2026 earnings, and Alphabet reported roughly $28.7 billion in similar unrealized gains.”

Such financial engineering, combined with circular financing schemes and inflated valuations, could very well set us up for an economy-wide crash. Or it could be even simpler. Rockefeller International chairman Ruchir Sharma believes the AI bubble could burst if the Fed decides to raise rates.

Ruchir Sharma believes monetary policy will cause the AI bubble to burst

Earlier this month, Sharma went on CNBC to discuss the AI bubble. Sharma, who has examined financial bubbles extensively, believes the current AI boom checks all four boxes of a bubble: overvaluation, over-investment, over-ownership, and over-leverage across the sector. Sharma notes that Meta, Amazon, and Microsoft also have recently become big issuers of debt, which he views as a classic late-cycle bubble indicator.

In Sharma’s estimation, the AI bubble burst will come down to monetary policy; he expects stock prices to go parabolic before the bubble pops, with higher interest rates (e.g., the 10-year Treasury yield going over 5%) eventually causing cheap capital to dry up. “Bubbles do not fall under their own weight. It is always higher interest rates that end big bubbles,” Sharma explained on CNBC.

So, what’s in store? An AI-induced tech bubble burst or an economy-wide crash

For those of us who foresee an AI-induced crisis and crash, there are several different possible outcomes. One is a correction reminiscent of the late 1990’s dot-com bubble, which resulted in a sectoral bubble burst in 2000 that was largely contained to Silicon Valley. In that crash, 200,000 people lost their jobs and thousands of tech companies went under. The U.S. economy did enter a brief and broader recession in March 2001; the stock market lost $8.3 trillion in value, making an impact in many 401(k)s.

That said, a tech bubble burst like the dot-com fiasco is arguably a better outcome than an economy-wide crash similar to the Great Financial Crisis of 2008. In that downturn, the housing crisis led to a full-blown recession that took down the entire economy. Unfortunately, this is a real possibility given our current economy’s overreliance on AI investments, circular financing, cross-holdings, and opaque financial engineering.

Sumit Sharma, an independent economist and tech policy expert, believes that antitrust enforcement from the FTC could potentially save us from an economy-wide crash. In an Op-Ed for Tech Policy Press, Sharma writes, “We need antitrust enforcement to ensure that AI firms compete independently and vigorously, without the contractual restraints of cloud-for-equity, without overlapping boards, and without minority stakes that quietly align incentives.”

Sharma sees a solution in the FTC’s January 2025 Staff Report on AI Partnerships and Investments. According to Sharma, the key motivation behind Microsoft’s investments in OpenAI was to create a moat around their cloud business. As Sharma explains, these cloud-for-equity deals “[bind] the most promising frontier-model developers to the very firms they might otherwise displace, while making it harder for cloud entrants and independent labs to compete on the merits. These contractual terms and cross-holdings simply create the appearance of competition while allowing consolidation.”

I agree with Sharma that unraveling these cross-holdings would be good for the average consumer and the overall economy in the long run. A correction is inevitable, and if the FTC unravels some of these cross-holdings, that could prevent one large AI company’s failure from spreading across an interlocked network.

Not everyone believes we’re in a bubble

I’d be remiss if I didn’t mention the folks who think we’re not in a bubble and headed for a serious correction. Many analysts at Goldman Sachs and J.P. Morgan believe we are not in an AI-fueled bubble. In a recent J.P. Morgan Asset Management report, investment specialist Nicholas Cangialosi and market strategist Stephanie Aliaga make the case that the hyperscalers’ large margins and strong cash flow make the current economic landscape different from the dot-com bubble and the 2008 recession.

Despite acknowledging the rampant circular financing between the frontier model developers, chip companies, and hyperscalers, Cangialosi and Aliga suggest that the hyperscalers’ balance sheets make them impervious to the tightening credit conditions that have burst past bubbles. Also, they point out that spending is primarily invested in physical AI infrastructure like chips and data centers.

They write, “At the peak of the dot-com era, only about 7% of the fiber-optic network was being utilized, leaving vast excess capacity that took years to absorb. But today, data center vacancy rates are at record lows and utilization levels hover around 80% […] Hyperscalers are already seeing returns through increased cloud demand and productivity gains in coding, advertising, and enterprise tools.” In my mind, this is an overly optimistic take.

Key takeaways

Whether you agree with me that we’re already in a bubble or not, the fact remains that policymakers should be prepared for a crash.

I agree that the cross-holdings among the largest frontier AI companies and members of the magnificent seven is distorting competition and could create a systemic crash.

It is also a huge red flag that OpenAI floated the idea of the U.S. government taking a 5% equity stake in their company earlier this month. OpenAI’s projected financial results foresee negative cash flow until 2030, and likewise, Anthropic isn’t expecting a profit for another four years at the earliest.

I agree with Asad Ramzanali and Ganesh Sitaraman at the Vanderbilt Policy Accelerator who call out the circular equity financing that is rampant in this space. They write, “In AI, vendors are taking equity stakes in unprofitable companies, at a scale that, based on original commitments, could include the largest-ever investment in a private company—because most of the money will be spent buying their own products.”

They are are right to bring attention to what they call an “extreme financialization of the AI sector, that includes financial engineering of various sorts, including circular equity investments, a wide array of debt vehicles that are complex and interlocking, and government subsidies, all of which create a financial picture that is opaque and shifts risk from companies to all of society.”

And when the AI bubble does burst, Ramzanali and Sitaraman argue that Congress should absolutely not bail out any AI firms, affected financial entities, and related tech companies. I could not agree more.

I don’t know if you’ve noticed, but voice AI is having a moment. ElevenLabs, leading AI voice generator, raised $500 million this year at an $11 billion valuation. The broader market pulled in $2.1 billion in VC funding in 2025, eightfold what it got the year before. By 2034, the voice AI agent market is expected to be worth close to $47.5 billion. So, it’s obvious that every major enterprise is either running a pilot, watching one, or being pitched one right now.

But while the industry has been busy celebrating its own growth, India did something in February that completely changes the strategy for anyone building or buying voice AI. It didn’t launch a startup or fund a model; instead, it published the foundation as a public good and invited everyone to build on it for free.

Sound familiar? It should, because India did exactly this with payment methods 10 years ago, and it rewrote the entire industry.

VoicERA and BHASHINI: India’s open-source voice AI infrastructure strategy

If you weren’t watching Indian FinTech closely in 2016, here’s a quick flashback. Before the Unifed Payments Interface (UPI), 90% of India’s transactions ran on cash. The digital alternatives that existed—bank transfers through NEFT and IMPS—charged per-transaction fees. Most small merchants simply didn’t bother with digital payments because the friction wasn’t worth it. Then, the government launched UPI: one open, shared payment layer, free for users and merchants, running in real time.

By 2025, it was processing 228 billion transactions worth $3.4 trillion a year, and handling nearly half of all real-time payment volume on the planet. The companies that had been profiting from the old infrastructure found themselves in an impossible spot. After all, you can’t compete with free.

Fast forward to February 2026. India held the India AI Impact Summit in New Delhi and launched VoicERA, an open-source voice AI stack running on BHASHINI, its national language platform. It handles real-time speech, conversational AI, and telephony across more than 700 dialects. It runs in the cloud or on-premises. And it was built from scratch, so no vendor can lock you in.

At the launch, BHASHINI CEO Amitabh Nag said: “In this era, India owns its voice.” That line is easy to read as a proud moment for a national initiative, but now read it again as a CFO. A government serving a billion people has just decided it will never pay foreign SaaS vendors by the minute for voice infrastructure. It built its own, put the code in the open, and handed it to every startup, enterprise, and government department to use.

The global voice AI world was too busy processing ElevenLabs’ valuation to notice. And in my opinion, that will be an expensive thing to have missed, because public infrastructure at this scale pulls pricing down across the entire category.

Why voice AI is harder to deploy than the demos suggest, and what India learned first

India’s relevance to this story goes beyond what any government announced.

The demos are good. Really good. The voice sounds human, the latency is near-instant, and the conversations feel natural. The problem surfaces the moment the demo ends—McKinsey’s 2025 State of AI research found that only 23% of organizations actually scale AI agents into production.

For voice AI specifically, that gap is even more punishing because a voice agent that works beautifully in a demo room meets a very different world in production—for example, accents the model wasn’t trained on; background noise; or customers who ramble, go off-script, or ask something the system was never designed to handle.

Gartner® puts the broader stakes bluntly: Over 40% of agentic AI projects will be cancelled outright by 2027 because the demo set expectations the deployment could never meet.

The model is almost never the reason things fall apart. The reason is everything the demo didn’t show, like the CRM integration that was never quite finished, the quality that degrades the moment real call volume hits, or the moment something breaks at 2am and nobody is sure whose problem it is.

India knows this intimately, because it never had the luxury of a controlled environment. United-States-trained speech models lose 15–25% accuracy on Indian English audio because of how acoustically different it is. More than 250 million Indians naturally switch between two languages mid-sentence—not occasionally, just as their normal way of talking. Building voice AI that works in those conditions means building something that can handle noise, surprise, and users who are not going to make life easy.

Voice AI use cases and ROI: What enterprise leaders should be building now

Here is the practical question worth sitting with if you’re an executive making a decision about voice AI.

The returns are real: Well-implemented systems have delivered 331%–391% ROI over three years. The bigger question is, what happens when every company in your industry has access to the same voice AI tools? So, ask yourself: When technology is no longer your advantage, what is?

In financial services, a voice agent can serve a customer your app never reached. For example, someone in a smaller city that’s more comfortable speaking than typing, wanting to talk in their own language about a loan or an account question.

In healthcare, voice makes it practical to check in with every discharged patient—something no call center team could scale to—catching early warning signs through natural conversation and feeding that directly into clinical records.

In B2B, voice handles the qualification, the scheduling, and the follow-up, freeing your senior people for the conversations that actually move deals, all without having to grow the team.

What makes any of this a lasting advantage has nothing to do with which platform you run it on. The platform will eventually become just another utility, like electricity or broadband. What compounds is the intelligence you build on top: what your system learns about your customers over time, the context it carries into every conversation, the data that shows you where things go wrong and why. That is the asset. That is what a competitor cannot pick up by switching vendors next year.

The strategic implications of India’s voice AI model for global enterprises

It’s easy to read the VoicERA story as something that only matters inside India—a government project solving a local problem. Except that’s not how infrastructure shifts work.

When VoicERA shows, at the scale of a billion users, that voice infrastructure can be open and publicly owned, charging premium SaaS prices for that same layer becomes hard to justify anywhere.

Picking the right vendor is the smallest decision in all of this. The harder questions are what you’re building on top of it, who owns the customer insights and conversation data it generates, and whether it holds up when real people use it. That’s the discipline India’s market forced onto its builders.

In early 2026, Block cut 40% of its workforce and told shareholders that the decision reflected how intelligence tools have changed what it means to build and run a company. Its stock rose 24% that day. Coinbase followed weeks later, cutting 700 jobs and telling employees the company needed to become AI-native; its stock gained too, according to CNBC’s coverage of the announcement.

Neither company was in financial distress. Both described themselves as profitable and growing. Framing the cuts around AI, rather than around softening demand or a correction to pandemic-era hiring, was a choice. It’s a choice more companies made in 2026 than in any year before it, whether or not AI was the actual reason the work still got done with fewer people.

That choice looks rational if the goal is a favorable market reaction rather than a verifiable one. For IT leaders heading into 2027 budget planning, the gap between the narrative markets currently pay for and the operational reality is becoming one of the more consequential accountability problems in enterprise technology.

The pattern behind 2026’s AI layoffs

Block and Coinbase are not outliers. Meta cut roughly 8,000 roles in May, which was about 10% of its workforce, in the same quarter it disclosed plans to spend $125-$145 billion on AI infrastructure. They reported quarterly revenue was up 33% year after year. None of these companies were in financial distress when they announced cuts, and all of them named AI as the reason.

Gartner’s own research complicates that explanation. A survey of 350 executives found that 80% of large enterprises piloting AI reported workforce reductions, but researcher Helen Poitevin found no correlation between those cuts and measurable ROI. Her analysis is clear: Workforce reductions can create budget room, but they do not create a return on their own

SHRM has a name for the space between the AI story a company tells and the AI outcome it can actually demonstrate: AI-washing, the practice of overstating AI’s role in a business decision. Outplacement firm Challenger, Gray & Christmas has tracked AI as the single most-cited reason for layoffs for four consecutive months in 2026.

What the market is actually paying for

Why would a market reward a story its own research firms can’t validate? Because a layoff framed around AI reads to investors as evidence of technological ambition rather than an admission of overhiring or soft demand, and markets have historically responded better to the former. Jason Schloetzer, a faculty affiliate at Georgetown University’s McDonough School of Business, offered SHRM a more direct explanation. Executives often cite AI when the real driver is that they lack the cash flow to fund AI investment without freeing up capital elsewhere. For investors, he noted, reduced expenditure reads as improved profit, regardless of the reason behind it.

Kenny Pyle, an HR technology analyst at SHRM, put the incentive plainly. Invoking AI lets a company send two positive signals in place of a negative one and implies it is technologically ahead of its peers, and that it is willing to make hard calls. Neither signal actually requires that an AI investment should replace anyone’s job.

What the AI efficiency story leaves out

On June 25, 2026, Apple raised prices across its Mac, iPad, Apple TV, HomePod, and Vision Pro lines up to$300 a unit, pointing to a global memory chip shortage that CEO Tim Cook called a hundred-year flood. Apple’s stock fell 6% that day, its steepest single-session drop since April 2025, erasing roughly $275 billion in market value. The increase was a plain admission that components now cost more, not a claim about headcount or productivity. The market punished that honest admission about as readily as it rewards the flattering layoff story.

The shortage behind Apple’s price increases is real, and it isn’t specific to Apple. AI is also making IT budgets more expensive industry-wide, for reasons that have nothing to do with any single company’s headcount decisions. IDC described the shortage as an unprecedented inflection point, severe enough that IDC expects it to persist well into 2027. Samsung, SK Hynix, and Micron, which together produce nearly all the world’s DRAM, have redirected manufacturing capacity toward the high-bandwidth memory used in AI data centers, because those chips carry substantially higher margins than the commodity memory used in laptops, phones, and enterprise servers. Deutsche Bank analysts, quoted in Fortune’s coverage of the shortage, called memory production a zero-sum game: Every wafer devoted to AI infrastructure is a wafer that does not become the RAM inside a piece of enterprise hardware.

That is the part of the AI cost story a headcount reduction cannot offset. If server and endpoint hardware costs are climbing because of a supply-side shock affecting the entire industry, cutting staff and calling it AI efficiency doesn’t make that shock disappear; it just changes which line item absorbs the pressure. IT leaders working through IT cost optimization frameworks are about to find memory and compute pricing behaving less like a controllable expense and more like a commodity, subject to swings no procurement strategy can fully insulate against.

Where the reckoning eventually lands

The stock market reward for an AI-framed layoff is captured immediately in the boardroom on the day of the announcement. The operational proof that AI actually delivered the value that was promised takes considerably longer to arrive, if at all, and it lands somewhere else entirely: It lands on the lean IT organization while absorbing rising infrastructure costs and delivering on commitments made in an earnings call.

Gartner has already put a number on how much of that promise won’t hold up. The firm expects more than 40% of agentic AI projects to be cancelled by the end of 2027, citing escalating costs, unclear business value, and inadequate risk controls. “Most agentic AI projects right now are early-stage experiments or proof of concepts that are mostly driven by hype,” said Anushree Verma, the Gartner analyst who led the research.

The same Gartner study that found no correlation between AI layoffs and AI ROI also identified what does correlate with it: Organizations seeing real returns are the ones investing in what Poitevin calls people amplification, training employees to build and run their own automations rather than treating headcount reduction as the source of value.

That’s the actual choice sitting in front of most IT leaders right now: Invest in people to get better at AI, or cut people and call it an AI outcome. The data already shows which one tends to work.

In this episode of the Thinking Tech series, Samudhra Senthil, enterprise evangelist at ManageEngine, explores a surprising paradox at the heart of modern innovation. Despite publishing more research than ever, the pace of groundbreaking discovery is actually slowing down. From the overwhelming flood of unverified information to the rise of AI tools reshaping how we think and work, this episode unpacks the forces stalling scientific progress. It asks whether artificial intelligence (AI) is the cure or part of the problem. Tune in for a thought-provoking look at how we can balance AI’s efficiency with the deep, curious thinking that drives true innovation.

What you’ll learn:

Why the pace of innovation is slowing down despite record levels of research output

How the complexity and volume of modern information is burying potentially disruptive ideas

What a study of 45 million papers reveals about the declining impact of new research

How AI tools like Elicit and Perplexity are changing the way researchers work

The risks of over-relying on AI, from distortions to the erosion of deep, original thinking

Why AI may be widening the innovation paradox, making iteration easier while stifling true breakthroughs

How to strike the right balance between AI efficiency and the human curiosity that drives real discovery

A few years ago, I fell down a rabbit hole researching how to scrub my digital footprint. What I found was discouraging. The services were expensive, and doing it myself would require a level of commitment I didn’t have the bandwidth for.

As it turns out, I’m among the 94% of Americans who’ve never used a data removal service, despite privacy concerns. A 2024 survey reveals reasons that echo my own: cost, skepticism, and confusion about where to start. It’s a question security teams are fielding more often, too, as employees and executives ask whether they’re worth the investment.

The promise of a clean slate is harder than ever to deliver in 2026. Services that once promised a digital “control-Z” have lost their luster as deletion looks less like erasure than endless maintenance. This article looks at what these services actually deliver, what they don’t, and what’s changed since they first made that promise.

What data removal services actually do

Data removal services first emerged in the 2010s to address a genuine problem. Data brokers and people-search sites were aggregating and selling personal information like addresses, phone numbers, and relatives to anyone willing to pay for it. Today the category has matured into a range of offerings, from automated broker opt-out tools to human-managed removal services and broader privacy bundles including VPNs and identity monitoring.

How these services work is, admittedly, a catch-22. To clear their digital footprint, users first have to hand their data over to a third party. And because brokers aren’t required to honor removal requests permanently, that data can reappear within months. That’s why these services are subscription-based, with some charging users up to $25 per month. Rather than erasing ourselves, we’re paying someone to follow behind us with a dustpan.

For some users, like executives and other high-visibility employees whose personal exposure raises spear-phishing and doxxing risk, these services offer real value. But for everyone else, I’m not so sure.

What they don’t cover

My lack of confidence in data removal services stems from the fact that they’re built for just one threat: the data broker ecosystem. A Consumer Reports investigation found these services largely ineffective at even that. Four months after the initial request, data brokers had removed only 35% of profiles. Some data removal services get as little as 4% to 6% of records removed.

But the broker ecosystem is only part of what our personal information exposes us to online, which also includes hacking, employer access to personal data, and legal discovery. Matthew “Dutch” Van Andel, a former Disney engineer with above-average personal opsec, encountered all three in quick succession. It cost him his job.

Despite using MFA and encrypted email, malware hidden in a GitHub plugin compromised Van Andel’s password manager and more than 1,000 accounts. Disney then accessed his personal iCloud through his work laptop and used his private browsing history in its investigation. When Van Andel sued, the process exposed him all over again.

Van Andel’s case isn’t an indictment of data removal services. But it is a reminder that our digital footprint is much larger than a name, address, and phone number on a people-search site. Employees should also know that anything stored, synced, or browsed on a work device is potentially accessible to their employer.

Managing expectations in 2026

Beyond their limitations, data removal services are up against a changing privacy landscape in the United States. California’s Delete Act centralizes deletion across over 500 registered brokers. Oregon, Texas, and Vermont require data brokers to register with the state and meet additional requirements. Living in Texas, it’s reassuring to know there’s at least some state-level protection in place, even if it falls short of what California offers.

Whether your state offers a centralized opt-out process or not, data removal services still help reduce exposure, just not to the extent some advertise. Questions to ask when evaluating a service include:

‐ Is it transparent about the data it finds as well as the measures used to protect my data? A service that won’t show what it found or how it handles your data is hard to trust.

‐ How many different data brokers do they cover? Most cover at least 100, but this figure can be difficult to verify. What matters more is whether the service can prove it actually found and removed your data.

‐ Will I be able to report sightings of my own data to be removed? Custom removals help catch misses and reappearances.

‐ Is this a standalone service, or does it come with other tools? Bundling VPN and antivirus can add value, but the removal function needs to be solid on its own.

‐ How easy do they make it to enter data? You’ll be submitting a lot of personal information, so the experience should be efficient. Errors for incomplete forms and redundant fields will slow you down.

Employees who’d rather handle data removal on their own should focus first on major people-search and data-broker sites before moving on to old accounts and apps. Advise them to keep records of every request, follow up in two to four weeks, and expect to repeat the process every three to six months in case data reappears.

Key takeaways

Anxiety about your digital footprint is understandable. But security teams should call data removal services what they are: a convenience like paying someone to clean your house or do your taxes. They’ll never be a substitute for MFA, strong password hygiene, or account monitoring, and organizations should continue to reinforce phishing awareness, social media policies, and other basic training.

India generates some of the largest volumes of personal data in the world. Every UPI transaction, every digital KYC verification, and every healthcare record leaves a trail of data moving across platforms, organizations, and infrastructure. For years, companies treated this as an asset to be maximized. The Digital Personal Data Protection Act (DPDPA) is forcing them to rethink.

Yet DPDPA compliance isn’t just a milestone to check off. It represents a fundamental shift in how organizations think about data ownership, accountability, and digital trust, and it raises harder questions than most businesses are prepared to answer.

To unpack what the DPDPA really means for Indian organizations, we’re joined by Anandaday Misshra, founder and managing partner of AMLEGALS, a pan-India law firm with over 27 years of experience across data governance, AI regulation, and international arbitration. He is also the architect of the Vibe Data Privacy™ framework, which helps enterprises build practical, scalable data governance systems.

In this conversation, he goes beyond the legislation to explore why consent is now a business process, why privacy is moving into the boardroom, and why organizations that treat the DPDPA as a check box exercise may be in for the most painful surprise when enforcement becomes real.

Agenda

What is driving India’s push for the DPDPA?
How the DPDPA compares to the GDPR
Consent is no longer just a check box
Shadow AI and the ₹250 crore compliance risk hiding in plain sight
What enforcement will realistically look like in India