TechCrunch+

Agriculture produces what might be some of the hardest planet-warming carbon emissions to eliminate. People need to eat, and right now, the way we make food generates a lot of greenhouse gases — about a third of the total created by human activity. There are some simple ways to reduce it, like eating less meat, but there’s a point where cutting is no longer an option.

That’s why there’s growing interest in something called enhanced rock weathering, where minerals that naturally absorb carbon dioxide from the atmosphere are crushed up and spread on soils to speed the process. It happens that some of these minerals are also beneficial for farmers and ranchers, many of whom have been searching in vain for ways to reduce their operations’ carbon emissions.

There are a few startups chasing this market, and one promising company, Eion, is announcing a $12 million Series A today, TechCrunch has exclusively learned. The round was led by AgFunder and Ridgeline, with participation from Carbon Removal Partners, Mercator Partners, Orion, Overture, SLVC, Trailhead Capital, and mineral supplier Sibelco, with which the company has an off-take agreement.

“We need a hockey-stick growth path to deal with the gigaton challenge that we have in front of us,” Eion founder and CEO Adam Wolf told TechCrunch. Enhanced rock weathering, he added, offers that potential.

Eion is one of the companies that Stripe’s climate program bought carbon credits from last year, and it’s been using that money to kick-start its operations. The new funding will allow the startup to finalize its methodology, expand operations and build a mill to crush rock to help produce its patented soil amendment called CarbonLock.

Stripe-backed Eion digs up $12M Series A to help farms capture carbon with green rock dust by Tim De Chant originally published on TechCrunch

Every developer knows that it’s a bad idea to hardcode security credentials into source code. Yet it happens and when it does, the consequences can be dire. Until now, GitHub only made its secret scanning service available to paying enterprise users who paid for GitHub Advanced Security, but starting today, the Microsoft-owned company is making its secrets scanning service available for all public GitHub repos for free.

In 2022 alone, the company notified partners in its secret scanning partner program of over 1.7 million potential secrets that were exposed in public repositories. The service scans repositories for over 200 known token formats and then alerts partners of potential leaks — and you can define your own regex patterns, too.

“With secret scanning we found a ton of important things to address,” said David Ross, a staff security engineer at Postmates. “On the AppSec side, it’s often the best way for us to get visibility into issues in the code.”

Now, if you host your code on GitHub, the company will automatically notify you directly about leaked secrets in your source code. This also means that you will get alerts for secrets where there isn’t a partner to notify (maybe because you self-host your HashiCorp Vault, for example).

To begin using the service, you have to enable the feature in their GitHub security settings. However, the rollout of the service will be gradual and it will not be available to all users until the end of January 2023.

GitHub’s own tool is, of course, not the only service that will scan for leaked secrets. There are also open-source tools like gitLeaks (which can integrate with GitHub actions) and a plethora of security companies like Nightfall and CheckPoint’s Spectral, though their services tend to go well beyond secret scanning and are generally geared toward enterprises.

GitHub brings free secret scanning to all public repos by Frederic Lardinois originally published on TechCrunch

Instagram is introducing and expanding a number of features to help users keep their accounts secure, the company announced on Thursday. Most notably, the social network is launching a new “hacked” hub where users can report and resolve account access issues. If you’re unable to log into your account, you can enter Instagram.com/hacked on your mobile phone or desktop browser to access the new hub.

Next, users will be able to select if they think they have been hacked, forgot their password, lost access to two-factor authentication or if their account has been disabled. From there, users will be able to follow a series of steps to regain access to their account. If a user has multiple accounts associated with their information, they will be able to choose which account needs support.

In addition to the roll out of this hub, Instagram is expanding access to a feature that is designed to give users multiple ways to get their account back if they lose access. Earlier this year, Instagram began testing a way for people to ask their friends to confirm their identity in order to help regain access to their account. Now, this option is available to everyone. If you are locked out of your account, you can choose two of your Instagram friends to verify your identity and get back into your account.

Instagram is also testing new ways to help prevent hacking on its platform before it actually happens. The company already removes accounts that its automated systems determine to be malicious, and is now taking this measure a step further.

“Because bad actors often don’t immediately use accounts maliciously, we’re now testing sending warnings if an account that we suspect may be impersonating someone requests to follow you,” the company wrote in a blog post. “In the coming months, we’ll also send warnings if an account that may be impersonating a business sends you a Direct Message (DM).”

The company is also going to start displaying the blue verified badge forverified accounts in more places across the platform to make it easier for users to quickly determine if the account they’re interacting with is legitimate. You can now see verified blue badges in Stories and DMs. Instagram will also start featuring them in the Feed in the future.

Instagram launches a new hub to help users resolve account access issues by Aisha Malik originally published on TechCrunch

The market for lab-grown meat, also called cultivated or cell-cultured meat, is expected to reach $1.99 billion by 2035, growing at an annual rate of 21.4%. Beef is poised to be the dominant segment.

The market got a boost last month when the U.S. Food and Drug Administration gave what amounts to a safety blessing to Upside Foods, a cultivated meat product startup, effectively setting in motion what many of these companies have been working toward: accelerated commercialization.

The FDA concluded that it had “no further questions” related to how Upside is producing its chicken made from the cultured cells of animals and said it is working with other cultivated meat companies in other pre-market consultation discussions.

However, cultivated meat continues to struggle with cost, chiefly how expensive it is to make products, which means that having price parity with traditional meat isn’t likely to happen soon. There’s also the all-important perspective of taste: Will people really want to eat these products?

While companies are working on taste, they are also making claims about the sustainability of the cultivated meat industry. A Good Food Institute report from last year showed that cultivated meat production processes could significantly reduce both global warming and land use. For beef, it can be, in some cases, reductions of more than 80% in environmental impact when compared to traditional beef production.

Today, cultivated meat startup SCiFi Foods, which raised $22 million this summer, revealed results from an analysis it conducted with The Ohio State University. It showed that 1 kilogram of its SCiFi burger had a smaller environmental impact than a traditional beef patty.

SCiFi’s burger consists of cultivated beef cells and plant-based ingredients, like water and soy protein isolate, and its production showed an overall greenhouse gas emissions reduction of 88.5%, while reducing energy use by 37.7%, land use by 90.6% and water use by 96.9%, according to a press release.

The company claims it is the first to have a study like this to prove its sustainability claims. SCiFi co-founder and CEO Joshua March told TechCrunch via email that “all previous studies were performed on generic, non-specific cultivated meat (pork, chicken, etc.). This is the first study that clearly lays out and quantifies the sustainable benefits of cultivated beef cells in detail. What makes it even more exciting is the potential for us to make our numbers even more impactful by using renewable sources of energy.”

Cultivated beef companies tout sustainability. Will it lead to marketability? by Christine Hall originally published on TechCrunch

While hiring freezes at Big Tech firms might be hurting certain AI investments, it’s clear that there remains a strong appetite throughout the enterprise for AI technologies — whether developed in-house or outsourced to third parties.

According to a McKinsey survey from early December, AI adoption at companies has more than doubled since 2017, with 63% of businesses expecting spending on AI to increase over the next three years. In February, IDC forecast that companies would increase their spend on AI solutions by 19.6% in 2022, reaching $432.8 billion by the end of the year and over $500 billion in 2023.

Generative AI is driving much of the recent corporate interest, with text-to-image tools such as OpenAI’s DALL-E 2 and Stable Diffusion seeing swift uptake despite the risks. Adobe just this month announced that it would open its stock image service, Adobe Stock, to creations made with the help of generative AI programs, following in the footsteps of Shutterstock (but not rival Getty Images). Meanwhile, Microsoft partnered with OpenAI to provide enterprise-tailored access to DALL-E 2 to customers like Mattel, which is using DALL-E 2 to come up with ideas for new Hot Wheels model cars.

Sequoia, the venture capital firm, said in a September blog post that it thought that generative AI could create “trillions of dollars of economic value.” That might sound optimistic, but there’s some evidence to suggest that AI has crossed the threshold from research project to serious revenue generator.

Companies — and VCs — continue to invest in AI despite market slowdown by Kyle Wiggers originally published on TechCrunch

Twitter is where the world shares its opinions and aspirations; it’s where brands, celebrities, and politicians interact with people, live and in the open.

From 2006 to 2012, Twitter’s public API was free for all, which developers took advantage of to build a wide range of value-add services (like TweetDeck) for the growing community. But after 2012, Twitter sharply curbed data access, eroding developer trust. To help reverse this trend and enable the developer community to flourish, Twitter acquired my startup, Reshuffle, in May 2021.

After the acquisition, Jack Dorsey and Bruce Falk (former CEO and GM of Revenue Products, respectively) charged me with re-opening the Twitter API. This came after I gave harsh feedback as an external developer in a public forum to Twitter’s leadership about how broken the developer platform was and the investment needed to correct it.

I told them that with proper attention, we could create a delightful and successful platform that gave developers the tools and APIs to thrive and improve Twitter’s user experience. Paraphrasing Ned Segal, Twitter’s then-CFO: “Amir told us how broken it is, so we bought his company to fix it.”

My startup and I joined an amazing, small team working to revive the Twitter developer platform. In addition to making sure developers had a growing suite of API endpoints that allowed them to build successful solutions, the team was also migrating the old API to new GraphQL-based infrastructure. On November 15, 2021, we officially launched the new Twitter V2 API, which was met with a lot of developer love and excitement.

But our ambition was bigger than that.

Dorsey and the board funded us even further (gave us approval to hire 50 more people) to build something much bigger — something they had been wanting to build for a long time. The vision was to make Twitter a true developer platform. We wanted developers to create in-Twitter apps that interact with users.

We envisioned a world in which you could share your favorite song from Spotify and listen to it live with all your followers on Twitter. We wanted you to be able to share your donation to your favorite cause and get your followers to donate as well through an integrated, GoFundMe-style experience. We wanted you to play Wordle inside Twitter, not just share the results. We wanted you to be able to interact with developer-powered apps inside the Twitter user experience.

That was just the beginning: we also envisioned a true decentralization of the Twitter timeline. We wanted to let developers create and share their own timelines.

Interested in tech? Here’s the TechCrunch-curated timeline. Interested in video games? Here is Twitch’s games and streamers timeline. Custom Timelines were the superhero evolution of Twitter Lists, giving developers advanced powers to curate and create their favorite topics, or do so on behalf of others.

The next part of our plan was to create a discovery mechanism, something like a store to discover and install these apps and timelines. We even started to explore the possibility to let developers monetize these experiences.

In the past year, we started to launch experiments around all these experiences.

We launched the tiles experiment, which was the first step towards apps:

We launched the timeline experiment, the first step towards open, custom timelines:

Twitter launched the toolbox, our first discovery experiment:

We pulled all that off during the rollercoaster of Twitter’s acquisition period. We believed (quite mistakenly) that Elon would spend time understanding the range of projects within Twitter and their impact on the public conversation. We believed that the developer platform was a crucial piece of his outspoken vision for an “everything app.” We were excited to present our vision to him, hopeful that he would be excited by our vision.

And then, on November 4, we were fired. Our work computers were bricked in the middle of the night and emails appeared in our personal accounts telling us we were fired.

According to one of our engineers’ public Tweets, two people remain out of our 100+ person organization. All our dreams and plans for developers were blown to dust.

When I joined Twitter, I’d told Dorsey, “We cannot mess this up. We can fix the relationship between developers and Twitter only once. If we blow this chance, I would not hire a developer that trusts our platform ever again.”

That’s because when developers start building on a platform, they’re making a bet that it will continue to exist with a high degree of stability. It’s a lot of work to build on a platform, and developers have been burned in the past by unsuccessful platforms such as Windows Mobile, and unreliable ones like some of the Facebook API.

I have worked on some of the best platforms out there — from Android and Sharepoint to Twitch and Slack — and they all have one thing in common: openness and trust.

Last month, we broke that trust, and I am sorry I couldn’t stop that from happening. I wake up in the middle of the night still thinking about it. We were excited and looking forward to announcing our vision to developers at Chirp last month, and now that vision is just an opening keynote document, lost on my bricked Twitter computer.

A developer once asked me how we could ensure that Twitter would continue to maintain and invest in this developer platform. My answer was, “As long as we have this amazing team our leadership tasked with building the platform, developers will see that we are serious about it, and I will let you know when that changes.”

Let this be my personal notice to Twitter developers: the team is gone, the investment has been undone. Love does not live here anymore.

The team that built the Twitter developer platform is amazing. They will build awesome platforms for developers and other developer tools in other companies. I am honored to have worked with each and every one of them.

As for me, as I move forward and transition into the venture capital world, I intend to invest in impactful developer platforms — ones that are committed to being open and trustworthy.

Developer platforms are all about trust, and Twitter lost it by Ram Iyer originally published on TechCrunch

Zipline got its start six years ago using its autonomous electric drones to deliver blood in Rwanda. Now, the logistics and drone delivery startup is expanding its Rwandan government partnership with a lofty aim: complete nearly 2 million instant deliveries and fly more than 200 million autonomous kilometers in the country by 2029.

The government of Rwanda and Zipline announced Thursday an expanded partnership that will add new delivery sites in rural and urban locations throughout the country — a move that is expected to triple its delivery volume.

The idea is to use Zipline to shore up Rwanda’s healthcare supply chain, address malnutrition and support the country’s eco-tourism industry, according to Rwanda Development Board CEO Clare Akamanzi, who touted this as a “national drone service.”

Deliveries will now include medicine, medical supplies, nutrition and animal health products. It also will give any government agency access to Zipline’s services, including the Ministry of Agriculture and Animal Resources, the Ministry of Information Communication Technology, the Rwanda Development Board, the Rwanda Medical Supply, and the National Child Development Agency.

The deal is validation for Zipline and could help it convince other countries to make similar nationwide partnerships. Zipline operates within Ghana, the U.S., Nigeria and Japan, and will be launching Côte D’Ivoire and Kenya soon, according to the company. This is the first time a government has tapped Zipline to provide a national drone service.

Zipline, which was founded in 2014, developed the entire ecosystem from the drones and logistics software to launch and landing system. It’s operations were limited in the beginning, starting in Rwanda with a focus on blood and vaccines and then expanding into Ghana. In the past year, the company has expanded its operational footprint and delivery volume — growth that was powered in part by $250 million in venture capital. (The company has raised $486 million to date.)

Zipline has delivered more than 450,000 packages to date with 215,000 deliveries occurring this year alone. The company has also snagged a number of partnerships in the past two years that signals aspirations to expand within and beyond healthcare. Zipline has partnerships with Toyota Group and UPS, it delivers medical equipment and personal protective gear for Novant Health in North Carolina and health and wellness products for Walmart.

Zipline is now the national drone service provider for Rwanda by Kirsten Korosec originally published on TechCrunch

The Linux Foundation today announced the launch of the Overture Maps Foundation, a new nonprofit organization that aims to enable developers to build new mapping products thanks to its interoperable map data and new tooling. The new organization was founded by AWS, Meta, Microsoft and TomTom, but the organization stresses that it is “open to all communities with a common interest in building open map data.”

When you think of open-source mapping, chances are that OpenStreetMap will be top of mind. After all, this project, too, is shepherded by a not-for-profit organization, with Microsoft and Meta also being sponsors of that project, too. Overture notes that it plans to integrate with existing projects like OpenStreetMap but also create new map data based on computer vision and other AI/ML techniques.

But it’s important to stress that the goal here is not to create a new map and compete with the likes of Google Maps or OpenStreetMap. It’s about the layer above that and enriching the existing base maps and allowing developers to build new products on top of them.

You’re not going to launch Overture Maps on your phone to get walking directions. Instead, the organization, for example, plans to define a common data schema and entity reference system for describing this data and run a quality assurance process over it to ensure that it can detect errors and potential data vandalism.

As Jim Zemlin, the Linux Foundation’s Executive Director told me, Overture fits into a larger trend of organizations that want to engage in open data initiatives to bring the same kind of open-source innovation they have seen in software to data initiatives.

“We’ve had several other open data initiatives, but this is one of the larger and I would say most ambitious open data initiatives that were bringing to the [Linux] Foundation,” said Zemlin. “We were approached by the founding companies of this initiative — so Meta, AWS, Microsoft and TomTom — who have an ambition to really create a set of open data for mapping that can allow for greater innovation in what they do, which is providing map services and geolocation data and so forth.”

It’s still very early days for this project. The Overture members are currently in the process of setting up the governance structure and the various committees that will make up their project. That also means that the group is only starting to define some of the data schemas and the kind of tooling it wants to develop.

“Microsoft is committed to closing the data divide and helping organizations of all sizes to realize the benefits of data as well as the new technologies it powers, including geospatial data,” said Russell Dicker, Corporate Vice President, Product, Maps and Local at Microsoft. “Current and next-generation map products require open map data built using AI that’s reliable, easy-to-use and interoperable.”

Microsoft, of course, has long offered Bing Maps (which mostly uses data from TomTom and OpenStreetMap for its base map), which you probably never use. The company’s most interesting use of Bing Maps is likely its Flight Simulator, though, which takes this base data and adds photogrammetry based on satellite imagery and procedurally generated buildings on top of the base map.

It’s also worth noting that TomTom, which you probably mostly remember from its stand-alone GPS hardware back in the day when that was still a thing, recently announced a new initiative to build a new map and development platform.

AWS may seem like the odd one out here, but the cloud giant has long had an interest in geospatial data, though it never offered its own maps and relied on partners like Esri and HERE Technologies. As Zemlin noted, the organization is already talking to a number of other potential partners but decided to launch with this initial group — in part, because at some point, you do have to make a project like this public to bring on additional members and build an ecosystem.

AWS, Meta, Microsoft and TomTom launch the Overture Maps Foundation by Frederic Lardinois originally published on TechCrunch

After Ross Chanin’s grandfather died, Chanin mourned not only him, but the fact that he’d never gotten a chance to hear more about his grandfather’s life. Over a conversation with a journalist friend, George Quraishi, it became clear to Chanin that Quraishi’s skill set — interviewing and audio editing — could be conducive to capturing a family’s history.

Chanin and Quraishi started conducting interviews for friends and family and recruited software engineers Martin Gouy and Moncef Biaz to build apps to make it easier to record remote interviews and play them back on the web. Convinced that they had the seeds of a business, Chanin and Quraishi decided to apply to Y Combinator and were accepted into the Summer 2020 batch.

Today, their startup — Artifact — has over 10,000 customers across 15 English-, Spanish- and French-speaking countries. It’s raised $5 million inclusive of a seed round led by GV, which had participation from Atento Capital, Goodwater and Offline Ventures and notable angels such as Y Combinator CEO Michael Seibel, Twitch CEO Emmett Shear and former Blizzard CEO Michael Morhaime.

“Interviews are incredible storytelling spaces, but they’re generally reserved for the rich and powerful and are not about our parents, grandparents and children,” Chanin told TechCrunch in an email interview. “Our dream is that Artifact will become the place where families the world over tell and experience their stories.”

Artifact charges customers $149 to have an interviewer (mostly moonlighting journalists, according to Chanin) conduct an interview with a family member. Packages include one interview and an edit with a custom introduction, sound mixing by an audio engineer and a web page for listening and adding photos.

It’s a four-step process. First, Artifact customers tell the interviewer who they’ll be interviewing and what they’ll discuss. Then, Artifact invites the interviewee to choose a day and time for the interview, which happens via phone or videoconferencing. The resulting recording — usually 30 minutes in length, give or take 15 minutes — is edited down to a 20-minute “episode,” which can be shared via the web with loved ones or publicly.

Artifact aims to turn around episodes within five business days of an interview. Up to two guests are included in the price of a single interview, with a $35-per-guest charge for additional interviewees.

“The people in your life may not be natural storytellers, but when they’re guided by professional interviewers, their stories become heirloom-quality episodes that live in your family’s private account,” Chanin said. “Once there, it’s easy to add photos and then securely share your Artifacts with the people you love.”

That’s a lot of sensitive info to upload to the cloud. But Chanin was adamant that Artifact doesn’t share personal data with third parties without “explicit and affirmative” consent from users. The platform stores data for as long as a person maintains an account, although users can delete recordings, notes and photos at any time.

Artifact is one startup among many delivering professional interviewing and audio biography services tailored to families. For instance, Vita’s app lets family members record audio stories and transcribe the text and even hand-select accompanying family photos, recipes and other media content for posterity if they so choose. Tales and Origin offer packages along the same vein while StoryWorth and StoryCorps are more self-service in nature, providing users with the tools to conduct interviews themselves including suggested lines of questioning.

So what sets Artifact apart? Chanin argues it “fills a need that remains unaddressed.”

“The large genealogy platforms do incredible work, helping families trace their lineage and build family trees. Cloud photo and print apps make experiencing family photos easy and fun. [But] while it’s one thing to trace your family’s history, it’s another entirely to record it in the voices of the people themselves,” Chanin said. “It’s the conversation spaces with our professional interviewers that create the magic — the intonations of voice, laughter and emotion that can make it feel like the person we’re listening to is sitting right there in the room with us.”

Artifact is also unique in that it operates on a marketplace model, connecting customers with freelance interviewers, audio editors and sound engineers who piece together each audio biography. The compensation structure for contractors wasn’t immediately clear in our interview with Chanin; we’ve asked Artifact for clarification.

In another differentiator, Artifact has dipped a toe into the corporate market, offering custom podcast creation services to companies, academic institutions and nonprofits. As with its biography business, Artifact’s enterprise-focused offering pairs customers with an interviewer who they instruct to talk to people about certain subjects, with Artifact handling all the scheduling, remote interviewing and editing.

To date, Artifact has produced podcasts for Clipboard Health, Onfleet, Yale, the University of Chicago and the Muscular Dystrophy Association, Chanin claims.

In a bid to remain ahead of rivals, Artifact aims to embrace merging AI technologies to further personalize the experience for its family biography customers. As customers upload photos and videos to their accounts, Artifact will soon begin marrying the images and videos to what’s being spoken about in an interview, Chanin says — no curation required.

“So, this is taking different types of media — image, video and text — and finding connections between them, then surfacing the result to the customer. We are calling this the ‘Sitback Experience,’ where users will simply click play, sit back and listen to people you love telling stories while relevant imagery and video play across the screen. It’ll be like a movie or a Ken Burns documentary about your family.”

Beyond the Sitback Experience, Artifact plans to launch Family Spaces, a dashboard where account holders will be able to add family members in a way that makes it clear which stories the platform’s recorded for individual people.

The swift development roadmap will keep Artifact a step beyond rivals, Chanin asserts, while delivering top-requested improvements to the user base. That’ll be key. Aside from the nascent enterprise venture, Artifact’s growth will depend on convincing existing customers to buy additional packages and new users to join in the first place.

“The pandemic reminded all of us that life is precious and that the people we love must never be taken for granted. In that way, Artifact provided a vehicle for many of our early adopters to act on those feelings and record family stories,” Chanin added. “From day one, we’ve built Artifact lean and as a service that provides immediate value to our customers — that our customers pay for. So in many respects, we launched the company the old fashioned way: introducing a new solution to a universal problem, learning from our customers and not focusing on growth at all costs.”

Chanin wouldn’t disclose Artifact’s burn rate. But he claimed that the company is “well capitalized,” with cash on hand for years. Artifact currently has a 14-person team (excepting the hundreds of freelancers in its marketplace) based in San Francisco and expects to “at least” double headcount over the next 12 months.

Artifact wants to record your family history in podcast-like audio recordings by Kyle Wiggers originally published on TechCrunch

Seeking to bring greater security to AI systems, Protect AI today raised $13.5 million in a seed-funding round co-led by Acrew Capital and Boldstart Ventures with participation from Knollwood Capital, Pelion Ventures and Aviso Ventures. Ian Swanson, the co-founder and CEO, said that the capital will be put toward product development and customer outreach as Protect AI emerges from stealth.

Protect AI claims to be one of the few security companies focused entirely on developing tools to defend AI systems and machine learning models from exploits. Its product suite aims to help developers identify and fix AI and machine learning security vulnerabilities at various stages of the machine learning life cycle, Swanson explains, including vulnerabilities that could expose sensitive data.

“As machine learning models usage grows exponentially in production use cases, we see AI builders needing products and solutions to make AI systems more secure, while recognizing the unique needs and threats surrounding machine learning code,” Swanson told TechCrunch in an email interview. “We have researched and uncovered unique exploits and provide tools to reduce risk inherent in [machine learning] pipelines.”

Swanson co-launched Protect AI with Daryan Dehghanpisheh and Badar Ahmed roughly a year ago. Swanson and Dehghanpisheh previously worked together at Amazon Web Services (AWS) on the AI and machine learning side of the business; Swanson was the worldwide leader at AWS’s AI customer solutions team and Dehghanpisheh was the global leader for machine learning solution architects. Ahmed became acquainted with Swanson while working at Swanson’s last startup, DataScience.com, which was acquired by Oracle in 2017. Ahmed and Swanson worked together at Oracle as well, where Swanson was the VP of AI and machine learning.

Protect AI’s first product, NB Defense, is designed to work within Jupyter Notebook, a digital notebook tool popular among data scientists within the AI community. (A 2018 GitHub analysis found that there were more than 2.5 million public Jupyter Notebooks in use at the time of the report’s publication, a number that’s almost certainly climbed since then.) NB Defense scans Jupyter notebooks for AI projects — which usually contain all the code, libraries and frameworks needed to train, run and test an AI system — for security risks and provides remediation suggestions.

What sort of problematic elements might an AI project notebook contain? Swanson suggests internal-use authentication tokens and other credentials, for one. NB Defense also looks for personally identifiable information (e.g., names and phone numbers) and open source code with a “nonpermissive” license that might prohibit it from being used in a commercial system.

Jupyter Notebooks are typically used as scratchpads rather than production environments, and most are locked safely away from prying eyes. According to an analysis by Dark Reading, fewer than 1% of the approximately 10,000 instances of Jupyter Notebook on the public web are configured for open access. But it’s true the exploits aren’t just theoretical. Last December, security firm Lightspin uncovered a method that could allow an attacker to run any code on a victim’s notebook across accounts on AWS SageMaker, Amazon’s fully managed machine learning service.

Other research firms, including Aqua Security, have found that improperly secured Jupyter Notebooks are vulnerable to Python-based ransomware and cryptocurrency mining attacks. In a 2020 Microsoft survey of businesses using AI, the majority said that they don’t have the right tools in place to secure their machine learning models.

It might be premature to sound the alarm bells. There’s no evidence that attacks are happening at scale, despite a Gartner report predicting an increase in AI cyberattacks through the end of this year. But Swanson makes the case that prevention is key.

“[Many] existing security code scanning solutions are not compatible with Jupyter notebooks. These vulnerabilities, and many more, are due to a lack of focus and innovation from current cybersecurity solution providers, and is the largest differentiation for Protect AI: Real threats and vulnerabilities that exist in AI systems, today,” Swanson said.

Beyond Jupyter Notebooks, Protect AI will work with common AI development tools, including Amazon SageMaker, Azure ML and Google Vertex AI Workbench, Swanson says. It’s available for free to start, with paid options to be introduced in the future.

“Machine learning is … complex and the pipelines delivering machine learning at scale create and multiply cybersecurity blind spots that evade current cybersecurity offerings, preventing important risks from being adequately understood and mitigated. Additionally, emerging compliance and regulatory frameworks continue to advance the need to harden AI systems’ data sources, models, and software supply chain to meet increased governance, risk management and compliance requirement,” Swanson continued. “Protect AI’s unique capabilities and deep expertise in the machine leaning lifecycle for enterprises and AI at scale helps enterprises of all sizes meet today’s and tomorrow’s unique, emerging and increasing requirements for a safer, more secure AI powered digital experience.”

That’s promising a lot. But Protect AI has the advantage of entering a market with relatively few direct competitors. Perhaps the closest is Resistant AI, which is developing AI systems to protect algorithms from automated attacks.

Protect AI, which is pre-revenue, isn’t revealing how many customers it has today. But Swanson claims that the company has secured “enterprises in the Fortune 500” across verticals, including finance, healthcare and life sciences, as well as energy, gaming, digital businesses and fintech.

“As we grow our customers, build partners and value chain participants we will use our funding to add additional team members in software development, engineering, security and go-to-market roles throughout 2023,” Swanson said, adding that Protect AI’s headcount stands at 15. “We have several years of cash runway available to continue to advance this field.”

Protect AI lands a $13.5M investment to harden AI projects from attack by Kyle Wiggers originally published on TechCrunch