Apple’s chief financial officer Luca Maestri challenged investor worries over an 8% drop in China revenue, by noting that sales in other emerging markets are growing.

“When we start looking at places like India, like Saudi, like Mexico, Turkey, Brazil…and Indonesia, the numbers are getting large, and we’re very happy because these are markets where our market share is [currenttly] low,” Maestri said Thursday during Apple’s second-quarter earnings call.

Revenue declined to $16.37 billion in China during the second quarter

“The populations are large and growing, and our products are really making a lot of progress within those markets,” continued Maestri. “The level of excitement for the brand is very high.”

One thing Maestri said there is verifiable: the populations in emerging markets are, in fact, large and growing. But Apple’s growth in those regions isn’t as rosy a picture as the executive attempted to paint, according to available data.

Net sales in the Americas — which would include places like Brazil and Mexico — were down slightly year-over-year from $37.8 billion to $37.3 billion, according to Apple’s Q2 2024 report. Sales in the “rest of Asia Pacific,” which would include emerging markets like India and Vietnam, were down 17% from $8.1 billion in the second-quarter of 2023 to $6.7 billion as of March 31.

To play devil’s advocate, Apple’s falling sales in those regions may have more to do with pricing than hype for the product.

Maestri noted that Apple has introduced several financing solutions and trade-in programs that “reduce the affordability threshold,” so that customers can buy in the top product range.

“That is very valuable for us in developed markets, but particularly in emerging markets where the affordability issues are more pronounced,” said Maestri.

Still, pointing to the beacon of hope that could be emerging markets may not be enough to settle down investors. China is Apple’s third-largest market, and it’s become a battleground of steep competition with domestic companies like Oppo and Xiaomi dominating the market. According to Counterpoint Research, Huwaei has has seen a massive swing in the country after being completely sidelined by U.S. sanctions. The firm’s phone sales increased almost 70% from the previous year, while Apple’s fell 19%. In September 2023, Beijing imposed bans on the iPhone for government officials in the workplace, echoing U.S. action against Huawei.

China and emerging markets aren’t the only downers on Apple’s balance sheet this quarter. The company also reported a 10% drop in iPhone sales across all markets. Apple’s slow adoption of AI versus competitors like Google and Microsoft have also potentially played a role in slowed down iPhone sales.

Despite unimpressive hardware figures, Apple still managed to beat Wall Street expectations. It also summoned a stock hike of more than 10% in after-hours trading, fueled by both an increase on services revenue and a massive $110 billion stock buyback — a jump over last year’s $90 billion purchase.

Investors on the call tried to get Maestri and Apple CEO Tim Cook to divulge some more details about its upcoming generative AI launches, which Apple has teased over the last few months, but the executives would only reveal that announcements were imminent.

We’ll be keeping our eyes out for Apple’s Worldwide Developer Conference for more news.

So far, the biggest surprise about May 7’s “Let Loose” event is that it’s happening at all. We’re just over a month out from Apple’s annual Worldwide Developer Conference, and yet the company determined there was enough news to warrant a stand-alone event. iPads (including the iPad Pro and iPad Air) will be the focus, while the long-neglected Apple Pencil is finally getting some love. It was, after all, the focal point of the event’s invite.

Rumors have also begun swirling around a potential M4 announcement. If true, it would represent a key change to Apple’s silicon release cadence, though that hasn’t exactly been consistent over the year. The likeliest reason for the change is Microsoft’s expected release of first-party, ARM-based silicon at its Build conference at the end of the month. Speculation has it that Microsoft’s AI-focused silicon will smoke Apple’s M3 chips.

It’s only been roughly seven months since Apple announced a trio of M3 chips. That’s a quick turnaround for a replacement chip, but keep in mind, the company is believed to have been working toward a more staggered release schedule last year, only to have those plans gunked up by some ongoing supply chain issues. Perhaps this schedule readjustment could see additional M4 SKUs announced during WWDC in a month.

In the meantime, it’s possible that, for the first time, the latest chip could appear on an iPad first — specifically the iPad Pro. At the very least, the high-end iPad is due for a refresh. The latest model, which sports the M2 chip, was released toward the end of 2022. Apple has spent the past several years blurring the line between the iPad and Mac, so why not let the tablet get its hands on new silicon first for a change?

The other big change to the Pro is said to be the long-awaited upgrade to an OLED display. Bloomberg suggests that this can’t come soon enough, as the tablet category is “in a deep funk right now.” Certainly the days of the iPad, one of Apple’s superstar devices, seem mostly behind us.

The company continues to dominate the category, but the tablet had a rough 2023, giving Apple a big slice of an increasingly shrinking pie. Canalys says the overall market decreased by 10% last year. Given the Vision Pro’s slow start, Apple could really use a shot in the arm right now — though a redesigned iPad doesn’t really fit the bill.

A pair of new iPad Airs is said to be arriving in 11- and 12.9-inch versions. The current rumor still has the more accessible models receiving an M2 chip — if the M4 does arrive next week, however, perhaps they’ll get an additional bump as well.

For what remains a relatively niche device, the Apple Pencil is apparently ready for its close-up. The stylus is said to be getting a new squeeze detection feature (which was alluded to earlier in an iPad) update. Haptic feedback could be on the docket as well, along with interchangeable magnet tips. Find My support could arrive too — a welcome addition for an eternally misplaced peripheral. Talking of blurring the lines between the iPad and Mac, Apple’s latest Magic Keyboard makes the tablet look even more like a laptop than before.

As for Macs themselves, with the arrival of the M4, I wouldn’t be wholly surprised to see the company sneak a computer refresh or two in. It seems just as — or even more — likely that Apple is saving major additions to the Mac line for WWDC in June.

The event is an early one for you West Coasters. It kicks off at 7 a.m. PT/10 a.m. ET. You can watch it live here and follow TechCrunch for the news as it breaks.

Microsoft has changed its policy to ban U.S. police departments from using generative AI for facial recognition through the Azure OpenAI Service, the company’s fully managed, enterprise-focused wrapper around OpenAI technologies.

Language added Wednesday to the terms of service for Azure OpenAI Service prohibits integrations with Azure OpenAI Service from being used “by or for” police departments for facial recognition in the U.S., including integrations with OpenAI’s text- and speech-analyzing models.

A separate new bullet point covers “any law enforcement globally,” and explicitly bars the use of “real-time facial recognition technology” on mobile cameras, like body cameras and dashcams, to attempt to identify a person in “uncontrolled, in-the-wild” environments.

The changes in terms come a week after Axon, a maker of tech and weapons products for military and law enforcement, announced a new product that leverages OpenAI’s GPT-4 generative text model to summarize audio from body cameras. Critics were quick to point out the potential pitfalls, like hallucinations (even the best generative AI models today invent facts) and racial biases introduced from the training data (which is especially concerning given that people of color are far more likely to be stopped by police than their white peers).

It’s unclear whether Axon was using GPT-4 via Azure OpenAI Service, and, if so, whether the updated policy was in response to Axon’s product launch. OpenAI had previously restricted the use of its models for facial recognition through its APIs. We’ve reached out to Axon, Microsoft and OpenAI and will update this post if we hear back.

The new terms leave wiggle room for Microsoft.

The complete ban on Azure OpenAI Service usage pertains only to U.S., not international, police. And it doesn’t cover facial recognition performed with stationary cameras in controlled environments, like a back office (although the terms prohibit any use of facial recognition by U.S. police).

That tracks with Microsoft’s and close partner OpenAI’s recent approach to AI-related law enforcement and defense contracts.

In January, reporting by Bloomberg revealed that OpenAI is working with the Pentagon on a number of projects including cybersecurity capabilities — a departure from the startup’s earlier ban on providing its AI to militaries. Elsewhere, Microsoft has pitched using OpenAI’s image generation tool, DALL-E, to help the Department of Defense (DoD) build software to execute military operations, per The Intercept.

Azure OpenAI Service became available in Microsoft’s Azure Government product in February, adding additional compliance and management features geared toward government agencies including law enforcement. In a blog post, Candice Ling, SVP of Microsoft’s government-focused division Microsoft Federal, pledged that Azure OpenAI Service would be “submitted for additional authorization” to the DoD for workloads supporting DoD missions.

Update: After publication, Microsoft said its original change to the terms of service contained an error, and in fact the ban applies only to facial recognition in the U.S. It is not a blanket ban on police departments using the service. 

 

Apple on Thursday reported a 10% drop in iPhone sales for the second fiscal quarter, dropping from $51.3 billion to $45.9 billion, year-over-year. The slowdown was fueled, in part, by an 8% drop in China.

Apple’s slow adoption of AI versus competitors like Google and Microsoft likely played a role in consumers’ decision to hold off on purchasing a new iPhone. Apple has promised some big announcements on that front (likely at WWDC in June), but the iPhone 16 itself likely won’t arrive until fall.

In spite of those dire hardware figures, however, the company still managed to beat Wall Street expectations, fueled by both an increase on services revenue and a massive $110 billion stock buyback — a jump over last year’s $90 billion purchase.

Services, which includes offerings like iCloud, Apple TV+ and Apple Music, jumped 14% for the year. Apple has long anticipated a slowdown in hardware sales, and its increasing focus on subscription services have helped to make up for some of that loss.

Developing…

Hubble Network has become the first company in history to establish a Bluetooth connection directly to a satellite — a critical technology validation for the company, potentially opening the door to connecting millions more devices anywhere in the world.

The Seattle-based startup launched its first two satellites to orbit on SpaceX’s Transporter-10 ride-share mission in March; since that time, the company confirmed that it has received signals from the onboard 3.5mm Bluetooth chips from over 600 kilometers away.

The sky is truly the limit for space-enabled Bluetooth devices: the startup says its technology can be used in markets including logistics, cattle tracking, smart collars for pets, GPS watches for kids, car inventory, construction sites, and soil temperature monitoring. Haro said the low-hanging fruit is those industries that are desperate for network coverage even once per day, like remote asset monitoring for the oil and gas industry. As the constellation scales, Hubble will turn its attention to sectors that may need more frequent updates, like soil monitoring, to continuous coverage use cases like fall monitoring for the elderly.

Once its up and running, a customer would simply need to integrate their devices’ chipsets with a piece of firmware to enable connection to Hubble’s network.

Hubble was founded in 2021 by Life360 co-founder Alex Haro, Iotera founder Ben Wild (who sold his startup to Ring), and aerospace engineer John Kim. Haro said the first time Wild presented the idea of connecting a Bluetooth chip to a satellite, his initial reaction was, “No freaking way.” And it does sound crazy, especially as consumer electronics can struggle to connect to other Bluetooth-enabled devices that are just a few feet away.

But the demand is there: existing IoT device are power hungry, are costly to operate, and lack global connectivity, the company says. These are fundamental limitations related to Bluetooth-enabled devices today, and they prevent many industries from leveraging IoT for their businesses.

The company joined Y Combinator’s Winter 2022 cohort and closed a $20 million Series A last March. Hubble’s first innovation was to develop software enabling off-the-shelf Bluetooth chips to communicate over very long ranges with low power.

On the space side, the company also patented a phased array antenna that can launch on a small satellite. The antennas work almost like a magnifying glass, and it’s what enables an off-the-shelf Bluetooth chip to communicate with the Hubble satellite. The team also had to solve Doppler-related problems, frequency mismatches occur between fast-moving objects exchanging data via radio waves.

Hubble is aiming to launch a third satellite on SpaceX’s Transporter-11 mission this summer and a fourth on Transporter-13. Those four satellites will compose what Haro called the “beta constellation,” and pilot customers are starting to turn their integrations on even today, he said. The startup plans to launch the following 32 satellites all at once in the fourth quarter of 2025 or the beginning of 2026, though the launch provider has not been selected yet.

Those 36 satellites will compose Hubble’s first “production constellation,” and they’ll enable connection with a Hubble satellite roughly 2-3 hours per day from anywhere in the world.

The graduates of Alliance DAO are often useful indicators of investor appetite and user adaption trends within the crypto space. The latest batch of the stage-agnostic crypto accelerator, unveiled today, comes at a moment of great excitement for the recovering market.

Just two months ago, Bitcoin hit its all-time high; though the value of the largest cryptocurrency has since declined, it continues to trade at much higher levels than during the market downturn following FTX’s implosion in late 2022. Venture investors are responding and plowing money into web3 startups, sending total fundraising in the space to roughly $1.9 billion in Q1, a sharp 58% jump from the quarter before, according to Crunchbase data.

The renewed enthusiasm among web3 believers is evidenced by catchphrases like “we are so back” that have filled crypto X/Twitter. In the meantime, regulatory efforts to rein in the industry have not waned. In the U.S., Binance’s Canadian founder Changpeng “CZ” Zhao is set to become the richest person to ever face imprisonment. Uniswap, which has been vocal about its decentralized approach to digital assets, received a notice from the U.S. Securities and Exchange Commission (SEC) last month.

Unsurprisingly, the ongoing crackdowns in the U.S. have a palpable impact on Alliance DAO’s geographic composition.

As shown in the graph (above) shared by Qiao Wang, one of the founding partners of Alliance DAO, founders based in North America accounted for 45% of the accelerator’s applicants in H2 2021; that share slipped to just 26% in H1 this year.

“Essentially, the U.S. is losing market share for crypto founders over the last three years. This is likely due to 1) regulations and 2) crypto finding product-market fit in emerging markets,” Wang told TechCrunch via email.

Indeed, the accelerator has seen a steady uptick in interest from Asia, which made up 24% of all applications in H1 2024, compared to 14% in H2 2021.

North America’s declining participation in Alliance DAO doesn’t imply founders simply abandon their crypto dreams. Historically, web3 entrepreneurs have been a flexible and nomadic tribe, fleeing crackdowns and seeking out more favorable regions. As a result, some of them may set up physical bases in emerging markets with a more amicable crypto environment.

As TechCrunch has reported, Asia has quickly become a popular destination for crypto entrepreneurs. The user base is large, young and open to new types of technology and financial assets. Several jurisdictions, including Hong Kong, Japan and Singapore, have taken notable steps to provide clearer regulatory frameworks for the budding sector, providing much reassurance to founders facing policy uncertainty elsewhere.

Meet the batch

Alliance DAO’s latest cohort, the 12th edition of its three-month program, received 1,503 applications. That marked a significant increase from the last batch’s 1083 applications. Just 21 teams were accepted this time, resulting in a competitive 1.4% acceptance rate. Twelve of them are presenting at today’s demo day.

Projects building on Ethereum, the most active blockchain by developer activity, are still the focus of this cohort, although other ecosystems like Solana and Bitcoin are “making a comeback,” according to Wang. Popular verticals seen across the batch include decentralized AI, crypto infrastructure (especially modular blockchain), decentralized finance (DeFi), and crypto-based payment solutions.

Now, let’s turn to the projects:

Company name: Villcaso

What it does: Permissionless U.S. real estate investing

Founders: Nathaniel Sokoll-Ward, Val Lee

The pitch: REITs, or real estate investment trusts, are designed to offer investors fractional exposure to real estate, lowering barriers to entry. While they offer more liquidity than traditional property investment, REITs are for the most part inaccessible to global investors, who make up an increasing share of total real estate investments in the U.S. Using a “fully legal permissionless token,” Villcaso is working to scale and distribute fractional ownership of U.S. real estate to a global audience. It has small equity positions in a large number of homes across the country.

Stage: Raising seed

 

Company name: GoBankless

What it does: Transferwise with stablecoins

Founders: Ygor Francisco, Khayalethu Mtshali

The pitch: GoBankless has its eye on Africa’s cross-border payments market that’s witnessing explosive growth. Businesses have been stuck with the long processing time and high settlement fees of traditional banks, while those that resist SWIFT’s monopoly are left facing counterparty risks in shadow markets. With the use of stablecoins, the startup is working to make cross-border payments instant without banking intermediaries. Today, GoBankless is serving around 50 small businesses across Mozambique and South Africa and settling $7 million in payments every month.

Stage: Raising seed

 

Company name: Wasabi Protocol

What it does: Leverage trading protocol

Founders: Eren Derman, Kemal Hasan Atay

The pitch: Crypto trading, especially longtail trading of new assets such as memecoins and NFTs, has seen a surge in daily volume. Popular platforms like Aevo and Hyperliquid allow users to gain early access, but they are “dependent on the market being sufficiently liquid,” leading to missed opportunities. Wasabi solves liquidity by backing user positions with underlying assets while its competitors take an algorithmic approach. Launched a few months ago, Wasabi’s total value locked (TVL) has grown to $60 million with over $200 million in volume.

Stage: Recently closed a seed round; raising a strategic round

 

Company name: Lulubit

What it does: Coinbase for Central America

Founders: Ianir Sonis, Diego Hernan Cabrera, Alan Futerman

The pitch: Central America is among regions that have shown a rapid pace of crypto adoption. Nonetheless, it’s still hard to even just buy and sell crypto in the region. P2P networks are unreliable while established exchanges charge high fees. Lulubit allows retail users in Central America to buy and sell crypto from their local banks and spend through the crypto debit card it issues; users can also send remittances on-chain to Lulubit and withdraw to their bank accounts at lower rates than the traditional method. Launched less than a year ago, Lulubit has amassed more than 18,000 users and processed over $1.3 million in volume in April alone, growing 36% month-over-month.

Stage: Raising seed

 

Company name: ZwapX

What it does: Marketplace for tokenized watches

Founders: Yohan Chiovetta, Noah Chiovetta, Rocco Di Capua

The pitch: The billion-dollar luxury watch market is enormous yet underserved by technological innovation. Peer-to-peer marketplaces are fraught with scams while B2C platforms face online authentication challenges. ZwapX offers a way for users to trade physical watches in the form of tokens, which act as certificates of ownership and authenticity. It has tokenized 44 watches to date with a $1.4 million TVL and a volume of $240,000.

Stage: Raising seed

 

Company name: Fractal Payments

What it does: Cross-border payments for global businesses

Founders: Pavel Skalin

The pitch: Money movement for businesses is one of the world’s biggest industries, yet it’s still suffering from perennial problems like high fees and slow processing. Fractal Payments is another startup aspiring to disrupt SWIFT with the use of stablecoins. Fully licensed in the European Union, it claims to make cross-border payments three times cheaper and six times faster than through legacy banking rails. It has facilitated more than $5 million in payments volume and working with a network of partners that support payments in over 60 countries.

Stage: Raising seed

 

Company name: Código

What it does: Crypto data for AI training

Founders: Jean-Philippe Emelie Marcos, Diego Besprosvan, Jaziel Guerrero

The pitch: Training data for AI is a billion-dollar market opportunity that has spawned unicorns like Scale AI. But existing solutions focus mostly on web2 use cases, with few powering AI training with crypto data. Código provides highly curated datasets to train specialized models for high-stake crypto applications, such as those involving financial transactions. Data is collected automatically through crowdsourcing, after which it is subject to a decentralized review and augmentation process where reviews can earn tokens. The tool has generated 4,000 dApps and four million lines of code within six months.

Stage: Raising seed

 

Company name: Accrue

What it does: Stablecoin payment network for Africa

Founders: Clinton Mbah, Adesuwa Omoruyi

The pitch: Bank transfers in Africa are notoriously costly and slow. Accrue aims to create a payment network that enables instant and affordable transactions — all powered by stablecoins. To that end, the startup is tapping the continent’s existing network of mobile tellers, which allow users to perform bank transactions over mobile phones, often simply through text messages. “10% of these mobile tellers are stablecoin-savvy,” and they are joining Accrue because it offers them more profit share and an upcoming token. The startup is cash-flow positive and has processed $5 million in payments.

Stage: Raising seed

 

Company name: Fig Investments

What it does: Tokenizing hedge fund strategies

Founders: Guanzhi Ma, Tony Qian

The pitch: The interest in decentralized finance (DeFi) services from traditional finance (TradFi) has surged, as seen in institutional players like Blackrock tokenizing stocks. Founded by banking veterans, Fig offers an automated trading desk that “matches TradFi interest in crypto with on-chain LP interest for returns.” It claims to be achieving a 10x scale than its competitor. Since launching four months ago, its TVL has grown to $10 million, with $40 million more in the backlog.

Stage: Raising seed

 

Company name: 0G

What it does: Modular AI chain

Founders: Michael Heinrich, Ming Wu

The pitch: 0G is building in the red-hot and cut-throat area of modular blockchain, which aims to help scale Ethereum transactions. Specifically, 0G is acting as a data availability layer, which ensures nodes in a blockchain network can access and verify transaction data. Its focus puts it in direct competition with well-funded projects such as a16z-backed EigenLayer, industry leader Celestia as well as Avail, which originated from Polygon. Using its unique technology, 0G claims it can achieve performance that’s 50,000 times faster than Celestia while costing 100x less than the rival.

Stage: Recently closed a 20x oversubscribed pre-seed round; raising seed

 

Company name: Proto

What it does: Google Maps on-chain

Founders: Akshay Yeleswarapu

The pitch: Despite the ubiquitous use of Google Maps, the application is surprisingly inaccurate in developing countries where cities are much denser than their Western counterparts and urban development happens rapidly. Proto wants to make navigation more accurate for underserved markets by crowdsourcing mapping data and allowing contributors to easily upload images with their mobile phones, a process incentivized by token rewards. Launched in late January, Proto has achieved 75% of Google Maps’ coverage of Bangalore through a network of 400 users.

Stage: Raising seed

 

Company name: Dinari

What it does: The global tokenized stock exchange

Founders: Gabriel Otte, Chas Rampenthal, Jake Timothy

The pitch: Global demand for U.S. securities has skyrocketed, yet access remains rather limited. Traditional brokerages have a high barrier of entry for foreign users, while early attempts to tokenize securities such as Ondo restrict certain features. Registered with the SEC, Dinari offers a way for non-U.S. investors to buy stocks via stablecoins. Its unique advantage is that its tokens are backed by real-world stocks. The platform’s TVL has grown to $500,000.

Stage: Closed a $10 million seed round; raising Series A

 

Alliance DAO invites a range of crypto experts to speak to cohorts about their domain knowledge. This time around, its guest mentors include Jacquelyn Melinek, founder of Token Relations and TechCrunch’s former crypto reporter; Jason Yanowitz, founder of Blockworks; Ming Ng, founder of Jupiter; Greg DiPrisco, founder of Ajna and M^0 Labs; Seung Yoon “SY” Lee, founder of Story Protocol; David Vorick, founder of Sia and Glow; and Ilja Moisejevs & Richard Wu, founders of Tensor.

Rivian is getting $827 million in incentives from the state of Illinois to support building its next-generation electric vehicle, known as the R2.

The company announced Thursday that the State of Illinois Department of Commerce and Economic Opportunity will dole out the funds to help expand its existing factory in Normal, IL, as well as build out supporting infrastructure and boost job training programs for Rivian’s workforce. Updates to the factory will begin “in the coming months.”

The funding announcement comes just two months after Rivian unveiled the mid-size R2 SUV, which is supposed to start at around $45,000 when it goes on sale in 2026. It also plans to make and sell a hatchback called the R3 that is powered by the same EV platform that underpins the R2.

Rivian had initially planned to build the R2 at a brand new $5 billion factory in Georgia. But the company announced at the R2 event in March that it had decided instead to build the R2 in Illinois, where it currently makes the R1T pickup, the R1S SUV, and its commercial electric vans.

The decision to pivot away from the Georgia plant allows the startup to move up the timeline for the R2 SUV, and also save more than $2 billion. Rivian has said it remains committed to building the factory in Georgia. But the company’s continued losses and struggles to scale production — it will make around the same number of EVs this year as it did last year — forced it to rethink its plans for the next-generation vehicle.

 

A controversial push by European Union lawmakers to legally require messaging platforms to scan citizens’ private communications for child sexual abuse material (CSAM) could lead to millions of false positives per day, hundreds of security and privacy experts warned in an open letter Thursday.

Concern over the EU proposal has been building since the Commission proposed the CSAM-scanning plan two years ago — with independent experts, lawmakers across the European Parliament and even the bloc’s own Data Protection Supervisor among those sounding the alarm.

The EU proposal would not only require messaging platforms that receive a CSAM detection order to scan for known CSAM; they would also have to use unspecified detection scanning technologies to try to pick up unknown CSAM and identify grooming activity as it’s taking place — leading to accusations of lawmakers indulging in magical thinking-levels of technosolutionism.

Critics argue the proposal asks the technologically impossible and will not achieve the stated aim of protecting children from abuse. Instead, they say, it will wreak havoc on Internet security and web users’ privacy by forcing platforms to deploy blanket surveillance of all their users in deploying risky, unproven technologies, such as client-side scanning.

Experts say there is no technology capable of achieving what the law demands without causing far more harm than good. Yet the EU is ploughing on regardless.

The latest open letter addresses amendments to the draft CSAM-scanning regulation recently proposed by the European Council which the signatories argue fail to address fundamental flaws with the plan.

Signatories to the letter — numbering 270 at the time of writing — include hundreds of academics, including well-known security experts such as professor Bruce Schneier of Harvard Kennedy School and Dr. Matthew D. Green of Johns Hopkins University, along with a handful of researchers working for tech companies such as IBM, Intel and Microsoft.

An earlier open letter (last July), signed by 465 academics, warned the detection technologies the legislation proposal hinges on forcing platforms to adopt are “deeply flawed and vulnerable to attacks”, and would lead to a significant weakening of the vital protections provided by end-to-end encrypted (E2EE) communications.

Little traction for counter-proposals

Last fall, MEPs in the European Parliament united to push back with a substantially revised approach — which would limit scanning to individuals and groups who are already suspected of child sexual abuse; limit it to known and unknown CSAM, removing the requirement to scan for grooming; and remove any risks to E2EE by limiting it to platforms that are not end-to-end-encrypted. But the European Council, the other co-legislative body involved in EU lawmaking, has yet to take a position on the matter, and where it lands will influence the final shape of the law.

The latest amendment on the table was put out by the Belgian Council presidency in March, which is leading discussions on behalf of representatives of EU Member States’ governments. But in the open letter the experts warn this proposal still fails to tackle fundamental flaws baked into the Commission approach, arguing that the revisions still create “unprecedented capabilities for surveillance and control of Internet users” and would “undermine… a secure digital future for our society and can have enormous consequences for democratic processes in Europe and beyond.”

Tweaks up for discussion in the amended Council proposal include a suggestion that detection orders can be more targeted by applying risk categorization and risk mitigation measures; and cybersecurity and encryption can be protected by ensuring platforms are not obliged to create access to decrypted data and by having detection technologies vetted. But the 270 experts suggest this amounts to fiddling around the edges of a security and privacy disaster.

From a “technical standpoint, to be effective, this new proposal will also completely undermine communications and systems security”, they warn. While relying on “flawed detection technology” to determine cases of interest in order for more targeted detection orders to be sent won’t reduce the risk of the law ushering in a dystopian era of “massive surveillance” of web users’ messages, in their analysis.

The letter also tackles a proposal by the Council to limit the risk of false positives by defining a “person of interest” as a user who has already shared CSAM or attempted to groom a child — which it’s envisaged would be done via an automated assessment; such as waiting for 1 hit for known CSAM or 2 for unknown CSAM/grooming before the user is officially detected as a suspect and reported to the EU Centre, which would handle CSAM reports.

Billions of users, millions of false positives

The experts warn this approach is still likely to lead to vast numbers of false alarms.

“The number of false positives due to detection errors is highly unlikely to be significantly reduced unless the number of repetitions is so large that the detection stops being effective. Given the large amount of messages sent in these platforms (in the order of billions), one can expect a very large amount of false alarms (in the order of millions),” they write, pointing out that the platforms likely to end up slapped with a detection order can have millions or even billions of users, such as Meta-owned WhatsApp.

“Given that there has not been any public information on the performance of the detectors that could be used in practice, let us imagine we would have a detector for CSAM and grooming, as stated in the proposal, with just a 0.1% False Positive rate (i.e., one in a thousand times, it incorrectly classifies non-CSAM as CSAM), which is much lower than any currently known detector.

“Given that WhatsApp users send 140 billion messages per day, even if only 1 in hundred would be a message tested by such detectors, there would be 1.4 million false positives every single day. To get the false positives down to the hundreds, statistically one would have to identify at least 5 repetitions using different, statistically independent images or detectors. And this is only for WhatsApp — if we consider other messaging platforms, including email, the number of necessary repetitions would grow significantly to the point of not effectively reducing the CSAM sharing capabilities.”

Another Council proposal to limit detection orders to messaging apps deemed “high-risk” is a useless revision, in the signatories’ view, as they argue it’ll likely still “indiscriminately affect a massive number of people”. Here they point out that only standard features, such as image sharing and text chat, are required for the exchange of CSAM — features that are widely supported by many service providers, meaning a high risk categorization will “undoubtedly impact many services.”

They also point out that adoption of E2EE is increasing, which they suggest will increase the likelihood of services that roll it out being categorized as high risk. “This number may further increase with the interoperability requirements introduced by the Digital Markets Act that will result in messages flowing between low-risk and high-risk services. As a result, almost all services could be classified as high risk,” they argue. (NB: Message interoperability is a core plank of the EU’s DMA.)

A backdoor for the backdoor

As for safeguarding encryption, the letter reiterates the message that security and privacy experts have been repeatedly yelling at lawmakers for years now: “Detection in end-to-end encrypted services by definition undermines encryption protection.”

“The new proposal has as one of its goals to ‘protect cyber security and encrypted data, while keeping services using end-to-end encryption within the scope of detection orders’. As we have explained before, this is an oxymoron,” they emphasize. “The protection given by end-to-end encryption implies that no one other than the intended recipient of a communication should be able to learn any information about the content of such communication. Enabling detection capabilities, whether for encrypted data or for data before it is encrypted, violates the very definition of confidentiality provided by end-to-end encryption.”

In recent weeks police chiefs across Europe have penned their own joint statement — raising concerns about the expansion of E2EE and calling for platforms to design their security systems in such as way that they can still identify illegal activity and send reports on message content to law enforcement.

The intervention is widely seen as an attempt to put pressure on lawmakers to pass laws like the CSAM-scanning regulation.

Police chiefs deny they’re calling for encryption to be backdoored but they haven’t explained exactly which technical solutions they do want platforms to adopt to enable the sought for “lawful access”. Squaring that circle puts a very wonky-shaped ball back in lawmakers’ court.

If the EU continues down the current road — so assuming the Council fails to change course, as MEPs have urged it to — the consequences will be “catastrophic”, the letter’s signatories go on to warn. “It sets a precedent for filtering the Internet, and prevents people from using some of the few tools available to protect their right to a private life in the digital space; it will have a chilling effect, in particular to teenagers who heavily rely on online services for their interactions. It will change how digital services are used around the world and is likely to negatively affect democracies across the globe.”

An EU source close to the Council was unable to provide insight on current discussions between Member States but noted there’s a working party meeting on May 8 where they confirmed the proposal for a regulation to combat child sexual abuse will be discussed.

Lacey Hunter thought all was well as she put her startup through the three-month Newchip accelerator. Then the organization filed for bankruptcy in May 2023. Things went from bad to worse later that year when she discovered warrants of her company — rights to buy an ownership stake — had become part of the proceedings, which ultimately forced her to shut down her company.

In 2022, Hunter started TechAid, an AI smart-matching tool for humanitarian aid, and was just beginning the accelerator’s curriculum when Newchip filed for bankruptcy.

“I made a few friends, but functionally, got nothing from Newchip,” Hunter said. “I was shooting to have the curriculum done by August, but in May, the website went down.”

The now-defunct Austin accelerator had filed for bankruptcy amid employee and customer discontent. The court has since ordered the company to auction off the warrants it held in more than 1,000 of the startups that went through the accelerator program.

Normally, private companies like startups have control over which investors are allowed to buy shares and the prices they pay. But the bankruptcy court, which works to restore creditors rather than equity holders, isn’t allowing Newchip’s startups to exert that kind of control. Instead, the auctions are ongoing, with the first tranche already sold and upcoming tranches expected to be sold this spring and summer.

Founders are outraged — including some, like Hunter, who have actually lost their companies as a result.

TechAid fought the sale of the warrants prior to closing the company. Hunter tried to buy them back herself from Newchip, but the organization’s lawyers declined her offer, she told TechCrunch. She had lined up a grant from a bank to help fund her offer, but it ultimately told her no because it was too risky for them to be involved with an unknown warrant holder on her cap table. So Hunter felt she had no choice but to shut TechAid.

“There was no path,” Hunter said. “I knew I was not going to be able to raise money. I mean, I couldn’t even get a no-strings-attached grant. I totally get that, but it still sucks.”

Newchip’s fall from accelerator grace

Newchip started out as an aggregator of top deals from “various equity-based crowdfunding platforms,” according to Silicon Hills News, and later evolved into an accelerator that promised to help startups grow their companies and meet investors — for a hefty fee.

It charged startups between a few thousand dollars and $18,000 to $20,000 for its training programs, founders said. Startups also granted Newchip the right to buy $250,000 worth of shares in the company at a later date, but at their current valuation — this type of deal is also known as a warrant.

Newchip founder and CEO Andrew Ryan previously faced harsh criticism about his leadership style, including allegations that he could be “abusive” and threatening to employees, according to eight former employees who walked out. (Ryan acknowledged to TechCrunch last year that his leadership style was based on “a military mindset.”) One example involved a meeting of about 15 employees in sales, operations and marketing. Ryan had asked the leaders of each department to read a book on how to help college volunteers be more passionate about volunteering, recalled one person who attended the meeting. Ryan asked two of the company’s leaders to lead the group in a discussion of the book. But many were confused by it and didn’t see how it applied to Newchip’s business.

“They were struggling with it. Andrew kept jumping in and interrupting them, and directly challenging them.” And finally, recalled the source, Ryan said, “This was a test for individuals that I’ve asked to do this today. I was going to fire one of you, based on whoever did the worst job.”

He then singled out one person, told the room the person was fired, and, this person recalled, Ryan then said, “I do stuff sometimes to see who’s loyal and to see who is going to do what I tell them to do. This was a test and you failed. You’re out.’”

After seeing Ryan fire this guy in front of the whole room, “I literally watched all of his direct reports sitting there saying to themselves, ‘I will never trust this man again,’” the source said.

Ryan contends that the person who was fired during that meeting had behaved aggressively after being singled out. Ryan also claims that the individual had come unprepared to lead the meeting, which Ryan viewed as an “act of overt insubordination,” telling TechCrunch: “While conducting the termination publicly in that meeting may seem harsh, it was intended to reinforce the gravity of the situation and ensure all managers understood that we took these training sessions and their responsibilities as leaders seriously.”

When Newchip (which also did business under the name Astralabs) initially filed for bankruptcy in March 2023, it was a Chapter 11 debt reorg. It then went into Chapter 7 — dissolution and liquidation —  two months later.

Its Chapter 11 filing revealed that it had $1.7 million in total assets and $4.8 million in total liabilities. But the value of the warrants was apparently not taken into account at that time, a source familiar with internal happenings said. Those warrants were estimated to be valued at an eye-popping just under $500 million by Austin-based VC fund and early Newchip investor Sputnik ATX, according to a document viewed by TechCrunch.

Management had not been keeping up with the warrants to the point where it had missed that some companies had exited or raised money, losing out on the potential upside, noted Kerstin Hadzik, a consultant who was brought in to serve as interim CFO a few weeks after the initial bankruptcy filing.

How much did Newchip potentially lose? Sputnik ATX said it identified $54 million in warrant value from companies that had liquidity events “that should have been reported to Newchip but were not,” according to documents viewed by TechCrunch.

In Hadzik’s view, Newchip might have also been saved from going into Chapter 7 if Ryan had been willing to step down as CEO and had presented the warrants as assets when initially filing for Chapter 11.

The judge repeatedly asked Ryan if he would voluntarily step down and let someone else, such as a chief restructuring officer, run the company. Ryan repeatedly dodged the question, expressing doubt that anyone could do so successfully. Ryan also noted that employees had requested “a new CEO” and later claimed that he “was going to step aside … but the shareholders and investors, as part of them putting capital in, preferred that I stay here to make sure that we have the capital … to continue driving the business.”

Ryan also admitted that he was the company’s “major owner and shareholder” and that he had just “terminated all the board” the week before, just after having filed for bankruptcy, according to court documents viewed by TechCrunch.

“The judge was offering like a lifeline,” and Ryan “just said no,” Hadzik recalled.

In a Zoom interview with TechCrunch back when we first reported on the bankruptcy, and in two LinkedIn posts in 2023, Ryan said that he accepted “full responsibility for the events at Newchip.”

Ryan later alleged that there was an attempted coup on the part of an investor but sources say that Ryan had actually asked early investor Joe Merrill to serve as CEO before changing his mind and resuming the role himself. Merrill, who was an early investor in Newchip under its previous model and also co-founder of Sputnik ATX, declined to comment beyond noting that he believed the attempted sale of the warrants was a valid move.

Founders fight for their companies

One founder, who asked to remain anonymous, told TechCrunch that Newchip had approached her on LinkedIn and told her if she got approved to join, she would get introductions to investors. So she paid a $7,500 deposit and was all set to join Newchip when a founder friend told her to “never pay for introductions.”

She decided to hear out Ryan. What convinced her to ask for her money back was that Ryan “blew off our meeting.” He reached out later, but she had already emailed Newchip asking for her deposit back on the basis that she had not started yet.

The founder got her money back, but Newchip didn’t void her contract, so she is now part of the bankruptcy lawsuit. That’s when she learned that someone could buy the warrants of her company for pennies on the dollar, and “it could screw your valuation going forward,” she said.

“I feel so much stress and embarrassment,” she told TechCrunch. “I’m a struggling founder and don’t have the money to pay for a lawyer. Here was this accelerator supposed to help founders, and instead it is imposing stress on young founders.”

There was a period of time when founders could object to their warrants being sold, according to Chad Harding, managing partner at Peak Technology Partners, the investment banking firm tasked by the court to sell the warrants.

The deadline for those in the first tranche to object to these sales was January 15, he told TechCrunch. Founders from all over the world, including Australia and Finland, filed objections, according to court documents.

“We were in the process of obtaining a refund from Newchip when Newchip went bust,” wrote Veronica Hey, CEO and founder of Australian startup Ok Away. “The contract is therefore null and void and the warrant attached to it is not applicable. None of this will stand up in an Australian court. If you continue to pursue in ‘selling’ this warrant you are selling something that does not exist and there will be repercussions.”

But startups’ objections were made in vain when the court overruled them. A bankruptcy court’s goal is to oversee the selling of assets to settle debts. If there is money left over, it’s paid to shareholders. Ryan is the majority shareholder.

So the warrants are being sold in three tranches. The first involved 133 companies, including for startups such as Cleanster.com, bitewell, Agshift and Firehawk Aerospace. Combined, those 133 startups had raised over $340 million in funding, according to documents shared by the sales agent with potential investors and viewed by TechCrunch.

Ultimately, the sales agent ended up selling 28 warrants in just four companies from the first tranche for a total of about $58,000, presumably at a discount. Successful bidders included Bitewell and ClearForce — startups that bought back their own warrants in advance for $5,000 each, according to an agreement with the trustee — as well as Palm Ventures and Angel Deal Syndicate. The latter purchased the bulk of the warrants, spending $43,000 on warrants in 24 companies, according to court documents viewed by TechCrunch.

The second tranche will likely be sold this summer and will include over 1,400 warrants for sale, according to Harding. The bid deadline will likely be late July, Harding said.

Founders of those startups included in the second tranche will also have the opportunity to object with a proposed deadline of May 31.

Ryan maintains that extensive efforts “have been made to notify stakeholders well in advance.”

“This has afforded ample time for interested parties to access information and documents, raise any objections or issues, and prepare for participation in the sale,” Ryan told TechCrunch.

When dreams become nightmares

Like TechAid’s Hunter, Garrett Temple blames the loss of his company on Newchip’s demise. He, similar to Hunter, also participated in Newchip’s accelerator program from January until May 2023. His startup, Novogiene, was a medical tech company focused on epidemic prevention.

Temple put around $7,500 on his credit cards to be part of the program and said that he never spoke with investors. His main reason for doing Newchip was to get investors for a $500,000 round, in part to pay for a small production run of his device so he could send it to universities and medical schools for pilot testing.

The meetings with investors were supposed to happen after a demo day that was scheduled for the summer. But when Newchip shut down in May, that demo day, and hence those introductions, didn’t happen. Temple wasn’t able to keep going and ended up dissolving Novogiene in the summer of 2023. As such, his company no longer existed for warrants to be sold to potential investors.

Temple said he spoke with his bank about getting money back from the program since he used credit cards. The bank was at first successful in getting $5,000 returned. However, about a month later, Temple noticed that money was no longer in his account and believes Newchip protested the funds.

Though Temple has moved on, he still has some intellectual property for Novogiene and says he is hoping at some point to license the technology to someone else or perhaps at another time pick up where he left off.

“It was very sad to call it quits because getting the funding to make those units was the only hurdle before making serious progress,” Temple said. “If they connected me with investors like they said, I could have made my invention, gotten efficacy and would be shipping units right now. I really do believe that.”

Accelerator operators sell dreams. But that doesn’t always mean that the accelerator will come through. And sadly, the founders who buy into those dreams can be the ones who end up paying the price.

AI startups face a different set of challenges from your typical SaaS company. That was the message from Rudina Seseri, founder and managing partner at Glasswing Ventures, last week at the TechCrunch Early Stage event in Boston.

Seseri made it clear that just because you connect to some AI APIs, it doesn’t make you an AI company. “And by AI-native I don’t mean you’re slapping a shiny wrapper with some call to OpenAI or Anthropic with a user interface that’s human-like and you’re an AI company,” Seseri said. “I mean when you truly have algorithms and data at the core and part of the value creation that you are delivering.”

Seseri says that means that there are major differences in how customers and investors judge an AI company versus a SaaS startup, and it’s important to understand the differences. For starters, you can put something that’s far from finished into the world with SaaS. You can’t do that with AI for a variety of reasons.

“Here’s the thing: With the SaaS product you code, you QA and you kind of get the beta — it’s not the finished product, but you can get it out there and can get going,” she said.

AI is a completely different animal: You can’t just put something out there and hope for the best. That’s because an AI product requires time for the model to get to a point where it is mature enough to work for actual customers and for them to trust it in a business context.

“In the early days, it’s a steep curve in learning and training the algorithm, and yet it has to be good enough for the customer to want to buy so it has to be good enough for you to create value,” she said. And that’s a hard line to find for an early stage startup.

And this makes it more challenging to find early adopters. She says you want to avoid the long call where the buyer is just trying to learn about AI. Startup founders don’t have time for calls like that. She says it’s important to focus on your product and help the buyer understand your value proposition, even if it’s not quite there yet.

“Always articulate the problem you are solving and what metric — how are you measuring it?” she said. Optimize on what matters to the buyer. “So you’re solving a problem that has business decision outcomes.” It’s OK to articulate your vision, but always be grounding your discussion in business priorities and how those are informing your algorithms.

How can AI startups win?

As you build your business, you need to be thinking about how you can stake a defensible place in AI, something that is particularly challenging as the big players continually carve out huge chunks of business ideas.

Seseri points out that in the cloud era, we had a foundation layer where the infrastructure players staked their claim; a middle layer where the platform players lived; and at the top we have the application layer where SaaS lived.

With the cloud, a few players like Amazon, Microsoft and Google emerged to control infrastructure. The foundation layer in AI is where the large language models live, and a few players like OpenAI and Anthropic have emerged. While you could argue these are startups, they aren’t in the true sense because they are being financed by those same big players who dominate the infrastructure market.

“If you’re going to compete for a new foundation layer, or you know, LLM play, it’s going to be very tough with multibillion dollar capital requirements, and at the end of the day, chances are it will end up being a commodity,” she said.

At the top of the stack is the application layer where thousands of SaaS companies were able to take advantage of in the cloud era. She said that the big players like Amazon, Google and Microsoft were not able to take all of the application layer business and there was room for startups to develop and grow into large, successful businesses.

There is also a middle layer where the plumbing gets done. She points to companies like Snowflake that have succeeded in building successful businesses in the middle layer by providing a place for the application players to put their data.

So where is she investing when it comes to AI? “I put my dollars in the application layer and very selectively in the middle layer. Because I think there is a moat around algorithms, whether it’s algorithms that are proprietary to you, or open source — and data. You don’t need to own the data. But if I have to pick, I’d like to have unique data access and unique algorithms. If I am forced to pick one, I will go after data,” she said.

Building an AI startup surely isn’t easy, perhaps even more challenging than a SaaS startup. But it’s where the future is, and companies that are going to try it have to know what they’re up against and build accordingly.