Twitter locked the Trump campaign out of its account for sharing COVID-19 misinformation

Twitter took action against the official Trump campaign Twitter account Wednesday, freezing @TeamTrump’s ability to tweet until it removed a video in which the president made misleading claims about the coronavirus. In the video clip, taken from a Wednesday morning Fox News interview, President Trump makes the unfounded assertion that children are “almost immune” from COVID-19.

“If you look at children, children are almost — and I would almost say definitely — but almost immune from this disease,” Trump said. “They don’t have a problem. They just don’t have a problem.”

While Trump’s main account @realDonaldTrump linked out to the @TeamTrump tweet in violation, it did not directly share it. In spite of some mistaken reports that Trump’s own account is locked, at this time his account had not been subject to the same enforcement action as the Trump campaign account, which appears to have regained its ability to tweet around 6PM PT.

“The @TeamTrump Tweet you referenced is in violation of the Twitter Rules on COVID-19 misinformation,” Twitter spokesperson Aly Pavela said in a statement provided to TechCrunch. “The account owner will be required to remove the Tweet before they can Tweet again.”

Facebook also took its own unprecedented action against President Trump’s account late Wednesday, removing the post for violating its rules against harmful false claims that any group is immune to the virus.

The president’s false claims were made in service of his belief that schools should reopen their classrooms in the fall. In June, Education Secretary Betsy DeVos made similar unscientific claims, arguing that children are “stoppers of the disease.”

In reality, the relationship between children and the virus is not yet well understood. While young children seem less prone to severe cases of COVID-19, the extent to which they contract and spread the virus isn’t yet known. In a new report examining transmission rates at a Georgia youth camp, the CDC observed that “children of all ages are susceptible to SARS-CoV-2 infection and, contrary to early reports, might play an important role in transmission.”

Gumroad founder Sahil Lavingia launches new seed fund in collaboration with AngelList

Gumroad founder Sahil Lavingia has teamed up with AngelList to launch his debut $5 million rolling fund to invest in early-stage entrepreneurs.

He is cutting $100,000 to $250,000 checks for startups and has a particular interest in B2B, SaaS, future of work, video, and developer tools. Limited partners include Arlan Hamilton, Josh Kopelman, and AngelList founder Naval Ravikant.

But, here’s the twist: Lavingia raised $5 million using just a Notion memo, a few tweets, and a Zoom call with over 1,800 registrants.

“It’s the power of Zoom and Twitter in the COVID era,” Lavingia said.

Still, two months ago, Lavingia didn’t even know he wanted to be a VC. The entrepreneur has made some angel investments in Lambda School, Figma, Haus, Clubhouse, and HelloSign (which was acquired by Dropbox). Eventually, though, he says angel investing got too expensive for him to do so he stopped.

Then, following George Floyd’s murder, he followed the lead of other investors rushing to invest in Black founders and tweeted this:

As a result of the tweet, he invested in 4 startups founded by Black entrepreneurs. Since some were looking on follow-on capital, he tapped into his network, including AngelList founder Naval Ravikant. Ravikant, seeing the deals, floated the concept of a rolling fund by him.

Rolling funds via Zoom

In February, AngelList launched a so-called rolling venture fund product to help emerging venture capitalists close their first funds, faster. The fund structure allows fund managers to raise new capital commitments on a regular basis and invest as they go, ergo the “rolling” aspect. Lavingia worked with AngelList to create his fund, and has capital commitments of $1.25 million per quarter in a $5 million per year fund.

The rolling fund structure can be a bit volatile because limited partners have to “re-up” their investments on a quarterly basis. It could put a fund’s investing ability in flux and thus impact portfolio construction, too.

One way to battle this volatility is that limited partners must commit to at least four consecutive quarters when investing in a rolling fund. After that, investors can choose on a quarter by quarter basis if they want to invest in the fund. Lavingia says that on this first close, he could have raised 5 to 10 times the capital, but chose to pick smaller checks from exceptional people. The smallest check in is $55,000 a year split over four quarters, he said.

Lavingia also claims that the rotating nature of check acceptance will allow him to continually invite a more diverse limited partner base as time goes on. He declined to share specific numbers on the current diversity of his LP base, but said that 30 percent of his portfolio companies to date are founded by Black entrepreneurs.

One other note on rolling funds, an SEC regulation — 506(c) — allows investors to publicly fundraise.Traditional venture capital funds are usually raised in private which disproportionately benefits those who already have their foot in the door. Lavingia says the 506(c) regulation allows him, as a first-time fund manager, to raise publicly on Zoom.

Lavingia hosted a Q&A about his new fund with a group of his buddies: Work Life Ventures’ Brianne Kimmel, AngelList’s Sunil Pai, and Earnest Capital’s Tyler Tringas.

Lavingia says that there were around 600 to 700 people live on the call, which is larger than most conferences he’s spoken at.

Lavingia was the second employee at Pinterest and left to start building Gumroad, a platform to help creators sell products to consumers. The company went through a gutting round of layoffs and restructuring in 2015, inspiring Lavingia to pen a viral blog post about his “failure to build a billion-dollar company.” Today, Gumroad is at $10 million ARR and is growing 100% year over year with a team of 10 people.

While Lavingia will continue to work on Gumroad, he says that his failure and transparency around it “is actually growing the company faster.”

‘I think it gives me a little bit more bandwidth to do an experiment along these lines,” he said, of the fund.

First-time fund managers have had to turn to unique ways to de-risk themselves in this volatile time. Lavingia’s story is no different, and showcases that the power of remote deals isn’t just a phenomenon that founders will benefit from.

As e-commerce accelerates, fintech startups raised record $100M rounds in Q2

Reading headlines here and there, one might assume that venture capital interest in fintech startups is setting records every quarter.

After all: didn’t Robinhood raise $280 million and $320 million more this year? Stripe raised $600 million just a few minutes ago, and wasn’t it Monzo that raised £60 million a few weeks back? Oh, and Hippo raised $150 million the other day.

And what about that huge Plaid exit earlier in the year and Chime’s jillion dollars that came right before 2019 ended?

That’s how it has felt to me, at least. And with good reason: new data from CB Insights indicates that fintech startups raised a record number of so-called “mega-rounds,” financings worth $100 million and more, in the second quarter of 2020.


The Exchange explores startups, markets and money. You can read it every morning on Extra Crunch, or get The Exchange newsletter every Saturday.


So the vibe in fintech that huge rounds have been landing quite often is correct. But underneath the big deals, there was early-stage weakness in the market that makes for a surprising contrast.

The same CB Insights report details a key “tailwind” factor for many fintech startups, namely that e-commerce is booming in the COVID-19 era, rising from about 16% of total U.S. commerce to around 27% through Q2 of this year.

So, let’s start by taking a quick look at Square’s earnings that leaked yesterday, and some notes from Shopify’s recent results to decipher just how fast the economy is heading online before examining what happened in Q2 VC for fintech startups as a cohort.

We’ll keep this as numbers-light as we can, and fun as we can — I promise. Let’s go!

Digital commerce is growing like a weed

You might think that Square, a company most famous for its IRL payment terminals and ability to turn any person into a micro-company would suffer while COVID-19 slowed in-person business. But, despite slowing gross payment volume (GPV), as expected, Square’s revenue exploded in Q2, growing from $1.17 billion in Q2 2019 to $1.92 billion in the most recent period.

What Q2 fundraising data tells us about the rest of 2020

It’s safe to say that no one could have predicted how this year’s fundraising marketplace was going to shape up. The beginning of the year saw us trending toward a blockbuster start, similar to 2018, rather than the steady burn of 2019. But after March there was no clear road map for how VCs and founders were going to react.

We’ve been tracking three key data metrics from the 2020 DocSend Startup Index to show us real-time trends in the fundraising marketplace. Using aggregate and anonymous data pulled from thousands of pitch deck interactions across the DocSend platform, we’re able to track the supply and demand in the marketplace, as well as the quality of pitch deck interactions.

The main two metrics are Pitch Deck Interest and Founder Links Created. These are leading indicators for how the fundraising marketplace is shaping up as it measures the activity happening around the pitch deck. As that interest peaks, we expect the amount of funds deployed to increase in the months after. Pitch Deck Interest is measured by the average number of pitch deck interactions for each founder happening on our platform per week, and is a great proxy for demand.

Founder Links Created is how many unique links a founder is creating to their deck each week; because each person you send a document to in DocSend gets a unique link, we can use this as a proxy for supply by looking at how many investors a founder is sharing their deck with per week.

Here’s what we saw in Q2 and how that will affect the rest of the year.

VCs are shopping

VC interest has been at an all-time high over the last quarter. Interest rebounded over the course of a few weeks after the pandemic was declared and shelter-in-place orders were given. But once interest rebounded to pre-pandemic levels it did something surprising. It kept climbing. In fact, the top 10 weeks for VC interest this year were all in Q2. Overall, interest was up 21.6% QoQ and 26% YoY. This means we’re looking at VCs viewing more pitch decks than they have any time in the last two years.

This is in spite of VC interest traditionally declining from late spring into summer, before bottoming out during the last two weeks of August. After the initial peak in the spring, VC interest typically doesn’t rebound until October.

But not only can we see that VCs are interacting with a lot of decks, we also can determine the quality of those interactions. We measure how long a VC spends reading each deck. From our previous research we know that the average pitch deck interaction is less than 3.5 minutes. But the amount of time VCs spent reading each deck in Q2 steadily declined, going below two minutes toward the end of the quarter. This tells us VCs are speeding through decks. That means they either know what they’re looking for and aren’t wasting time, or they’re scrutinizing decks less, opting for a Zoom call to hear more from a founder.

For founders, this means having a tight deck is even more important than before. Don’t have more than 20 slides, don’t send your appendix in your send-ahead deck and keep your slides concise and thoughtful (read our guide on how to put together a send-ahead deck here).

If you’re still not able to get a meeting with a VC during this intense shopping season, you may want to consider changing your fundraising strategy.

Founder timelines have changed

We can see over the last quarter that there have been clear spikes in the amount of links founders are sending out. Founders sent out 11% more deck links in Q2 than they did in Q1, but what’s interesting is that the number of links created actually dropped below 2019 levels on three separate occasions. So while founders might have been rushing to send their deck out during unstable times, there were plenty of weeks where founders were hanging back.

This conflicting story can tell us several things. First, founders have most likely condensed their fundraising efforts. According to our research earlier this year, the average pre-seed round takes longer than three months to complete. For those fundraising during a pandemic, three months can seem like a lifetime. This is not only due to the logistics of setting meetings with VCs who have packed calendars, but also the iteration process of receiving feedback from a potential investor, working on your deck, then sending it out to new targets. With global uncertainty, many founders likely decided to shorten their time away from their business by reducing their fundraising efforts to just a few weeks.

Second, due to aggressive cost cutting at the beginning of the pandemic, many founders found themselves with more runway than they expected. In fact, according to a recent survey we did, nearly 50% of founders changed their fundraising timeline by either moving it forward or delaying it. Founders that could afford to decided to avoid the volatile fundraising marketplace in an effort to preserve their valuations.

We’re looking at more than displaced interest from March

While it was easy during April and early May to think the fundraising marketplace was experiencing delayed activity due to the crash in March, the sustained interest makes it hard to believe that’s still the case, especially taking into account seasonality. The last week of the quarter saw a 37% increase in interest over 2019 and an 18% increase over 2018. With that level of activity, we’ve clearly entered a new normal for fundraising.

While valuations might be fluctuating, it’s quite clear VCs are shopping. To figure out why, you don’t have to look any further than the 2008 financial crisis. The businesses born out of crises tend to address real, systemic problems that require big, bold fixes. And the pandemic has certainly laid bare many societal issues that are worth addressing.

What Q3 and Q4 could look like based on current trends

If it’s clear that VCs are shopping, and it’s clear that this isn’t displaced interest from earlier this year, what does that mean for the future? We would normally see an increase in founder activity starting in late summer, leading to peak VC interest in the fall. Founder activity has been up and down, and VC interest has been steadily rising, which tells us there’s still pent-up demand to deploy capital. We should also see many founders who delayed their fundraising efforts enter the marketplace in the next few months. If pandemic conditions worsen, we might also see founders who had decided to push their fundraising efforts to next year moving their timelines forward.

If the current level of interest represents the new normal for VCs, we expect it to only increase as we enter the fall. And with more founders coming online in early to late fall, that pent-up demand should result in an increasingly active market. If you’re a founder, I would recommend kicking off your fundraise now in order to capitalize on the increased interest from investors and decreased competition for at least the first pitch meeting.

Q3 2020 is primed to be an intense shopping season for VCs

With the high possibility of an extremely active fundraising marketplace for the rest of the year, founders need to know how to take advantage of it. As you can see from the DocSend Pitch Deck Interest Metrics, spikes in the marketplace previously have resulted in some pretty specific behaviors by VCs.

Here are some tips on how to use the increasing levels of VC interest to your advantage.

VCs are spending less time on your deck, so get to the point

We’re seeing record low time spent per pitch deck. We know from previous research that VCs spend on average 3.5 minutes per pitch deck. But over the last quarter that time has dipped below three minutes. That can actually be a good and a bad thing. It implies that VCs are streamlining their process of looking at decks, which means they most likely know what they want. The downside of this is if you break a few cardinal rules right now your deck could end up in the reject pile.

From our research, VCs expect a deck to be around 20 pages. They expect a straightforward narrative that starts with your problem, leading to the solution, and then your product and business model. Our data found that VCs respond best to 35-50 words per slide (too few words per slide is also an issue; you want to offer enough context for your deck to make sense without you presenting it). The only place you can increase your word count is on your Team page. Our data shows the average number of words on a successful Team slide is 80. This gives you room to highlight the founding team’s relevant experience and show how you’re uniquely suited to build your business.

You have to include a “why now” slide and it should mention COVID-19

We already know that investors respond well to a Why Now slide. Our research shows that 54% of successful pitch decks included a Why Now slide, where only 38% of failed decks included it. That slide now has to work twice as hard. We’re hearing from investors that they expect to see information in your pitch deck about how your business has been affected by COVID-19 and how you plan to manage that impact moving forward. Even if the pandemic has had no material effect on your business, the investor will still have the question. Get out in front of it with a well-formed response near the beginning of your deck.

Amid pandemic, returning to offices remains an open question for tech leaders

As COVID-19 infections surge in parts of the U.S., many workplaces remain empty or are operating with skeleton crews.

Most agree that the decision to return to the office should involve a combination of business, government and medical officials and scientists who have a deep understanding of COVID-19 and infectious disease in general. The exact timing will depend on many factors, including the government’s willingness to open up, the experts’ view of current conditions, business leadership’s tolerance for risk (or how reasonable it is to run the business remotely), where your business happens to be and the current conditions there.

That doesn’t mean every business that can open will, but if and when they get a green light, they can at least begin bringing some percentage of employees back. But what that could look like is clouded in great uncertainty around commutes, office population density and distancing, the use of elevators, how much you can reasonably deep clean, what it could mean to have a mask on for eight hours a day, and many other factors.

To get a sense of how tech companies are looking at this, we spoke to a number of executives to get their perspective. Most couldn’t see returning to the office beyond a small percentage of employees this year. But to get a more complete picture, we also spoke to a physician specializing in infectious diseases and a government official to get their perspectives on the matter.

Taking it slowly

While there are some guidelines out there to help companies, most of the executives we spoke to found that while they missed in-person interactions, they were happy to take things slow and were more worried about putting staff at risk than being in a hurry to return to normal operations.

Iman Abuzeid, CEO and co-founder at Incredible Health, a startup that helps hospitals find and hire nurses, said her company was half-remote even before COVID-19 hit, but since then, the team is now completely remote. Whenever San Francisco’s mayor gives the go-ahead, she says she will reopen the office, but the company’s 30 employees will have the option to keep working remotely.

She points out that for some employees, working at home has proven very challenging. “I do want to highlight two groups that are pretty important that need to be highlighted in this narrative. First, we have employees with very young kids, and the schools are closed so working at home forever or even for the rest of this year is not really an option, and then the second group is employees who are in smaller apartments, and they’ve got roommates and it’s not comfortable to work at home,” Abuzeid explained.

Those folks will need to go to the office whenever that’s allowed, she said. For Lindsay Grenawalt, chief people officer at Cockroach Labs, an 80-person database startup in NYC, said there has to be a highly compelling reason to bring people back to the office at this point.

Facebook fights order to globally block accounts linked to Brazilian election meddling

Facebook has branded a legal order to globally block a number of Brazilian accounts linked to the spread of political disinformation targeting the country’s 2018 election as “extreme”, claiming it poses a threat to freedom of expression outside the country.

The tech giant is simultaneously complying with the block order — beginning Saturday after it was fined by a Supreme Court judge for non-compliance — citing the risk of criminal liability for a local employee were it not to do so.

However it is appealing to the Supreme Court to try to overturn the order.

A spokesperson for the tech giant sent us this statement on the matter:

Facebook complied with the order of blocking these accounts in Brazil by restricting the ability for the target Pages and Profiles to be seen from IP locations in Brazil. People from IP locations in Brazil were not capable of seeing these Pages and Profiles even if the targets had changed their IP location. This new legal order is extreme, posing a threat to freedom of expression outside of Brazil’s jurisdiction and conflicting with laws and jurisdictions worldwide. Given the threat of criminal liability to a local employee, at this point we see no other alternative than complying with the decision by blocking the accounts globally, while we appeal to the Supreme Court.

On Friday a judge ordered Facebook to pay a 1.92 million reais (~$367k) fine for non compliance, per Reuters, which says the company had been facing further daily fines of 100,000 reais (~$19k) had it not applied a global block.

Before the fine was announced Facebook had said it would appeal the global block order, adding that while it respects the laws of countries where it operates “Brazilian law recognizes the limits of its jurisdiction”.

Reuters reports that the accounts in question were controlled by supporters of the Brazilian president, Jair Bolsonaro, and had been implicated in the spread of political disinformation during the country’s 2018 election with the aim of boosting support for the right wing populist.

Last month the news agency reported Facebook had suspended a network of social media accounts used to spread divisive political messages online which the company had linked to employees of Bolsonaro and two of his sons.

In a blog post at the time, Facebook’s head of security policy, Nathaniel Gleicher, wrote: “Although the people behind this activity attempted to conceal their identities and coordination, our investigation found links to individuals associated with the Social Liberal Party and some of the employees of the offices of Anderson Moraes, Alana Passos, Eduardo Bolsonaro, Flavio Bolsonaro and Jair Bolsonaro.”

In all Facebook said it removed 33 Facebook accounts, 14 Pages, 1 Group and 37 Instagram accounts that it identified as involved in the “coordinated inauthentic behavior”.

It also disclosed that around 883,000 accounts followed one or more of the offending Pages; while the Group had around 350 accounts signed up; and 918,000 people followed one or more of the Instagram accounts.

The political disops effort had spent around $1,500 on Facebook ads, paid for in Brazilian reais, per its account of the investigation.

Facebook said it had identified a network of “clusters” of “connected activity”, with those involved using duplicate and fake accounts to “evade enforcement, create fictitious personas posing as reporters, post content, and manage Pages masquerading as news outlets”.

An example of removed content that was being spread by the disops network identified by Facebook (Image credit: Facebook)

The network posted about “local news and events including domestic politics and elections, political memes, criticism of the political opposition, media organizations and journalists”; and, more recently, about the coronavirus pandemic, it added.

In May a judge in Brazil had ordered Facebook to a block a number of accounts belonging to Bolsonaro supporters who had been implicated in the election meddling. But Facebook only applied the block in Brazil — hence the court order for a global block.

While the tech giant was willing to remove access to the inauthentic content locally, after it had identified a laundry list of policy contraventions, it’s taking a ‘speech’ stance over purging the fake content and associated accounts internationally — arguing such an order risks overreach that could damage freedom of expression online.

The unstated implication is authoritarian states or less progressive regimes could seek to use similar orders to force platforms to apply national laws which prohibit content that’s legal and freely available elsewhere to force it to be taken down in another jurisdiction.

That said, it’s not entirely clear in this specific case why Facebook would not simply bring down its own banhammer on accounts that it has found to have so flagrantly violated its own policies on coordinated authentic behavior. But the company has at times treated political ‘speech’ as somehow exempt from its usual content standards — leading to operating policies that tie themselves in contradictory nots.

Its blog post further notes that some of the content posted by the Brazilian election interference operation had previously been taken down for violating its Community Standards, including hate speech.

The case doesn’t just affect Facebook. In May, Twitter was also ordered to block a number of accounts linked to the probe into political disops. It’s not clear what action Twitter is taking.

We’ve reached out to the company for comment.

AI is struggling to adjust to 2020

2020 has made every industry reimagine how to move forward in light of COVID-19: civil rights movements, an election year and countless other big news moments. On a human level, we’ve had to adjust to a new way of living. We’ve started to accept these changes and figure out how to live our lives under these new pandemic rules. While humans settle in, AI is struggling to keep up.

The issue with AI training in 2020 is that, all of a sudden, we’ve changed our social and cultural norms. The truths that we have taught these algorithms are often no longer actually true. With visual AI specifically, we’re asking it to immediately interpret the new way we live with updated context that it doesn’t have yet.

Algorithms are still adjusting to new visual queues and trying to understand how to accurately identify them. As visual AI catches up, we also need a renewed importance on routine updates in the AI training process so inaccurate training datasets and preexisting open-source models can be corrected.

Computer vision models are struggling to appropriately tag depictions of the new scenes or situations we find ourselves in during the COVID-19 era. Categories have shifted. For example, say there’s an image of a father working at home while his son is playing. AI is still categorizing it as “leisure” or “relaxation.” It is not identifying this as ‘”work” or “office,” despite the fact that working with your kids next to you is the very common reality for many families during this time.

Image Credits: Westend61/Getty Images

On a more technical level, we physically have different pixel depictions of our world. At Getty Images, we’ve been training AI to “see.” This means algorithms can identify images and categorize them based on the pixel makeup of that image and decide what it includes. Rapidly changing how we go about our daily lives means that we’re also shifting what a category or tag (such as “cleaning”) entails.

Think of it this way — cleaning may now include wiping down surfaces that already visually appear clean. Algorithms have been previously taught that to depict cleaning, there needs to be a mess. Now, this looks very different. Our systems have to be retrained to account for these redefined category parameters.

This relates on a smaller scale as well. Someone could be grabbing a door knob with a small wipe or cleaning their steering wheel while sitting in their car. What was once a trivial detail now holds importance as people try to stay safe. We need to catch these small nuances so it’s tagged appropriately. Then AI can start to understand our world in 2020 and produce accurate outputs.

Image Credits: Chee Gin Tan/Getty Images

Another issue for AI right now is that machine learning algorithms are still trying to understand how to identify and categorize faces with masks. Faces are being detected as solely the top half of the face, or as two faces — one with the mask and a second of only the eyes. This creates inconsistencies and inhibits accurate usage of face detection models.

One path forward is to retrain algorithms to perform better when given solely the top portion of the face (above the mask). The mask problem is similar to classic face detection challenges such as someone wearing sunglasses or detecting the face of someone in profile. Now masks are commonplace as well.

Image Credits: Rodger Shija/EyeEm/Getty Images

What this shows us is that computer vision models still have a long way to go before truly being able to “see” in our ever-evolving social landscape. The way to counter this is to build robust datasets. Then, we can train computer vision models to account for the myriad different ways a face may be obstructed or covered.

At this point, we’re expanding the parameters of what the algorithm sees as a face — be it a person wearing a mask at a grocery store, a nurse wearing a mask as part of their day-to-day job or a person covering their face for religious reasons.

As we create the content needed to build these robust datasets, we should be aware of potentially increased unintentional bias. While some bias will always exist within AI, we now see imbalanced datasets depicting our new normal. For example, we are seeing more images of white people wearing masks than other ethnicities.

This may be the result of strict stay-at-home orders where photographers have limited access to communities other than their own and are unable to diversify their subjects. It may be due to the ethnicity of the photographers choosing to shoot this subject matter. Or, due to the level of impact COVID-19 has had on different regions. Regardless of the reason, having this imbalance will lead to algorithms being able to more accurately detect a white person wearing a mask than any other race or ethnicity.

Data scientists and those who build products with models have an increased responsibility to check for the accuracy of models in light of shifts in social norms. Routine checks and updates to training data and models are key to ensuring quality and robustness of models — now more than ever. If outputs are inaccurate, data scientists can quickly identify them and course correct.

It’s also worth mentioning that our current way of living is here to stay for the foreseeable future. Because of this, we must be cautious about the open-source datasets we’re leveraging for training purposes. Datasets that can be altered, should. Open-source models that cannot be altered need to have a disclaimer so it’s clear what projects might be negatively impacted from the outdated training data.

Identifying the new context we’re asking the system to understand is the first step toward moving visual AI forward. Then we need more content. More depictions of the world around us — and the diverse perspectives of it. As we’re amassing this new content, take stock of new potential biases and ways to retrain existing open-source datasets. We all have to monitor for inconsistencies and inaccuracies. Persistence and dedication to retraining computer vision models is how we’ll bring AI into 2020.

First US apps based on Google and Apple Exposure Notification System expected in ‘coming weeks’

Google Vice President of Engineering Dave Burke provided an update about the Exposure Notifications System (ENS) that Google developed in partnership with Apple as a way to help public health authorities supplement contact-tracing efforts with a connected solution that preserves privacy while alerting people of potential exposure to confirmed cases of COVID-19. In the update, Burke notes that the company expects “to see the first set of these apps roll out in the coming weeks” in the U.S., which may be a tacit response to some critics who have pointed out that we haven’t seen much in the way of actual products being built on the technology that was launched in May.

Burke writes that 20 states and territories across the U.S. are currently “exploring” apps that make use of the ENS system, and that together those represent nearly half (45%) of the overall American populace. He also shared recent updates and improvements made to both the Exposure Notification API, as well as to its surrounding documentation and information that the companies have shared in order to answer questions state health agencies have had, and hopefully make its use and privacy implications more transparent.

The ENS API now supports exposure notifications between countries, which Burke says is a feature added based on nations that have already launched apps based on the tech (that includes Canada, as of today, as well as some European nations). It’s also now better at using Bluetooth values specific to a wider range of devices to improve nearby device detection accuracy. He also says that they’ve improved the reliability for both apps and debugging tools for those working on development, which should help public health authorities and their developer partners more easily build apps that actually use ENS.

Burke continues that there’s been feedback from developers that they’d like more detail about how ENS works under the covers, and so they’ve published public-facing guides that direct health authorities about test verification server creation, code revealing its underlying workings, and information about what data is actually collected (in a de-identified manner) to allow for much more transparent debugging and verification of proper app functioning.

Google also explains why it requires that an Android device’s location setting be turned on to use Exposure Notifications – even though apps built using the API are explicitly forbidden from also collecting location data. Basically, it’s a legacy requirement that Google is removing in Android 11, which is set to be released soon. In the meantime, however, Burke says that even with location services turned off, no app that uses the ENS will actually be able to see or receive any location data.

Mammoth Biosciences’s CRISPR-based COVID-19 test receives NIH fundings through RADx program

CRISPR tech startup Mammoth Biosciences is among the companies that revealed backing from the National Institutes of Health (NIH) Rapid Accleration of Diagnostics (RADx) program on Friday. Mammoth received a contract to scale up its CRISPR-based SARS-CoV-3 diagnostic test in order to help address the testing shortages across the U.S.

Mammoth’s CRISPR-based approach could potentially offer a significant solution to current testing bottlenecks, because it’s a very different kind of test when compared to existing methods based on PCR technology. The startup has also enlisted the help of pharma giant GSK to develop and produce a new COVID-19 testing solution, which will be a handheld, disposable test that can offer results in as little as 20 minutes, on site.

While that test is still ind development, the RADx funding received through this funding will be used to scale manufacturing of the company’s DETECTR platform for distribution and use in commercial laboratory settings. This will still offer a “multi-fold increase in testing capacity,” the company says, even though it’s a lab-based solution instead of a point-of-care test like the one it’s seeking to create with GSK.

Already, UCSF has received an Emergency Use Authorization (EUA) from the FDA to use the DETECTR reagent set to test for the presence of SARS-CoV-2, and the startup hopes to be able to extend similar testing capacity to other labs across the U.S.