TechCrunch+

Special episode: Augmenting creativity with Alice Albrecht from re:collect (Found) by Theresa Loconsolo originally published on TechCrunch

Data breaches can be extremely harmful to organizations of all shapes and sizes – but it’s how these companies react to the incident that can deal their final blow. While we’ve seen some excellent examples of how companies should respond to data breaches over the past year — kudos to Red Cross and Amnesty for their transparency — 2022 has been a year-long lesson in how not to respond to a data breach.

Here is a look back at this year’s badly handled data breaches:

Nvidia

Chipmaker giant Nvidia confirmed it was investigating a so-called “cyber incident” in February, which it later confirmed was a data extortion event. The company refused to say much else about the incident, and, when pressed by TechCrunch, declined to say how it was compromised, what data was stolen, or how many customers or employees were impacted.

While Nvidia stayed tight-lipped, the now-notorious Lapsus$ gang quickly took responsibility for the breach and claimed it stole one terabyte of information, including “highly confidential” data and proprietary source code. According to data breach monitoring website Have I Been Pwned, the hackers stole the credentials of more than 71,000 Nvidia employees, including email addresses and Windows password hashes.

DoorDash

In August, DoorDash approached TechCrunch with an offer to exclusively report on a data breach that exposed DoorDash customers’ personal data. Not only is it unusual to be offered news of an undisclosed breach before it’s announced, it was even stranger to have the company decline to answer nearly every question about the news it wanted us to break.

The food delivery giant confirmed to TechCrunch that attackers accessed the names, email addresses, delivery addresses, and phone numbers of DoorDash customers, along with partial payment card information for a smaller subset of users. It also confirmed that for DoorDash delivery drivers, or Dashers, hackers accessed data that “primarily included name and phone number or email address.”

But DoorDash declined to tell TechCrunch how many users were affected by the incident — or even how many users it currently has. DoorDash also said that the breach was caused by a third-party vendor, but declined to name the vendor when asked by TechCrunch, nor would it say when it discovered that it was compromised.

Samsung

Hours before a long July 4 holiday, Samsung quietly dropped notice that its U.S. systems were breached weeks earlier and that hackers had stolen customers’ personal information. In its barebones breach notice, Samsung confirmed unspecified “demographic” data, which likely included customers’ precise geolocation data, browsing and other device data from customers’ Samsung phones and smart TVs, was also taken.

Now at year’s end, Samsung still hasn’t said anything further about its hack. Instead of using the time to draft a blog post that says which, or even how many customers are affected, Samsung used the weeks prior to its disclosure to draw up and push out a new mandatory privacy policy on the very same day of its breach disclosure, allowing Samsung to use customers’ precise geolocation for advertising and marketing.

Because that was Samsung’s priority, obviously.

Revolut

Fintech startup Revolut in September confirmed it was hit by a “highly targeted cyberattack”, and told TechCrunch at the time that an “unauthorized third party” had obtained access to the details of a small percentage (0.16%) of customers “for a short period of time.”

However, Revolut wouldn’t say exactly how many customers were affected. Its website says the company has approximately 20 million customers; 0.16% would translate to about 32,000 customers. However, according to Revolut’s breach disclosure, the company says 50,150 customers were impacted by the breach, including 20,687 customers in the European Economic Area and 379 Lithuanian citizens.

The company also declined to say what types of data were accessed. In a message sent to affected customers, the company said that “no card details, PINs or passwords were accessed.” However, Revolut’s data breach disclosure states that hackers likely accessed partial card payment data, along with customers’ names, addresses, email addresses, and phone numbers.

NHS supplier Advanced

Advanced, an IT service provider for the U.K.’s NHS, confirmed in October that attackers stole data from its systems during an August ransomware attack. The incident downed a number of the organization’s services, including its Adastra patient management system, which helps non-emergency call handlers dispatch ambulances and helps doctors access patient records, and Carenotes, which is used by mental health trusts for patient information.

While Advanced shared with TechCrunch that its incident responders — Microsoft and Mandiant — had identified LockBit 3.0 as the malware used in the attack, the company declined to say whether patient data had been accessed. The company admitted that “some data” pertaining to over a dozen NHS trusts was “copied and exfiltrated,” but refused to say how many patients were potentially impacted or what types of data were stolen.

Advanced said there is “no evidence” to suggest that the data in question exists elsewhere outside our control and “the likelihood of harm to individuals is low.” When reached by TechCrunch, Advanced chief operating officer Simon Short declined to say if patient data is affected or whether Advanced has the technical means, such as logs, to detect if data was exfiltrated.

Twilio

In October, U.S. messaging giant Twilio confirmed it was hit by a second breach that saw cybercriminals access customer contact information. News of the breach, which was carried out by the same “0ktapus” hackers that compromised Twilio in August, was buried in an update to a lengthy incident report and contained few details about the nature of the breach and the impact on customers.

Twilio spokesperson Laurelle Remzi declined to confirm the number of customers impacted by the June breach or share a copy of the notice that the company claims to have sent to those affected. Remzi also declined to say why Twilio took four months to publicly disclose the incident.

Rackspace

Enterprise cloud computing giant Rackspace was hit by a ransomware attack on December 2, leaving thousands of customers worldwide without access to their data including archived email, contacts, and calendar items. Rackspace received widespread criticism over its response for saying little about the incident or its efforts to restore the data.

In one of the company’s first updates, published on December 6, Rackspace said that it had not yet determined “what, if any, data was affected,” adding that if sensitive information was affected, it would “notify customers as appropriate.” We’re now at the end of December and customers are in the dark about whether their sensitive information was stolen.

LastPass

And finally, but by no means the least: The beleaguered password manager giant LastPass confirmed three days before Christmas that hackers had stolen the keys to its kingdom and exfiltrated customers’ encrypted password vaults weeks earlier. The breach is about as damaging as it gets for the 33 million customers who use LastPass, whose encrypted password vaults are only as secure as the customer master passwords used to lock them.

But LastPass’ handling of the breach drew a swift rebuke and fierce criticism from the security community, not least because LastPass said that there was no action for customers to take. Yet, based on a parsed read of its data breach notice, LastPass knew that customers’ encrypted password vaults could have been stolen as early as November after the company confirmed its cloud storage was accessed using a set of employee’s cloud storage keys stolen during an earlier breach in August but which the company hadn’t revoked.

The fault and blame is squarely with LastPass for its breach, but its handling was egregiously bad form. Will the company survive? Maybe. But in its atrocious handling of its data breach, LastPass has sealed its reputation.

It’s all in the (lack of) details: 2022’s badly handled data breaches by Carly Page originally published on TechCrunch

2022 was the year that electric vehicles entered the mainstream. Not everyone has one, but buying an EV no longer makes you an outlier. Driven by policy initiatives from governments and billions of dollars in investment from automakers, we can safely say the EV industry has begun to take shape.

Over the next year, that landscape will develop beyond the foundations of 2022. Here are some of our best guesses for what you can expect.

There will be a race to sell U.S.-built EVs in the first quarter

The Inflation Reduction Act, which the Biden administration passed in August, has already had a huge effect on the EV industry as automakers work to onshore their supply chains and factories. But with certain aspects of the IRA’s EV tax credit rules now to be delayed until March 2023, we’re expecting to see EV sales take off in the first quarter of the year.

Under the bill, eligible EVs could qualify for a $7,500 tax credit if they meet the requirements of being built in North America and having sourced critical battery materials from the U.S. or free trade agreement countries. Those rules were meant to go into effect on January 1, 2023, but the Treasury Department has delayed guidance on the critical materials rule until March. And it’s a good thing, too. While automakers in 2022 scrambled to set up factories in the U.S., most critical materials still come from China, so they need time (likely years) to set up new supply chains.

The delay means that a whole host of North American-built cars will now be eligible for the full refund, at least for the first three months of the year. The biggest winners will probably be Tesla and General Motors, whose sales caps under the previous EV tax incentives will be waived in the new year. But others like Ford, Nissan, Rivian and Volkswagen have all got a lineup of NA-built EVs that are ready to reap the benefits.

Even more EV models and sales

Electric vehicle sales in 2022 were pretty much dominated by who you’d expect: Tesla’s Models S, Y and 3, Chevrolet’s Bolt and Ford’s Mustang Mach-E. In the backdrop, nearly every automaker, be they a legacy OEM or a startup, unveiled a slew of impressive EVs for the 2023 market, from the Alfa Romeo Tonale to the Indi One. Most of them were geared towards the luxury consumer, though. In the next year, we’ll seeeven more new models come outthat are priced much more affordably.

In addition, expect the sheer number of new EVs on the market to pick up as new factories come online.McKinsey predicts legacy automakers and EV startups will produce up to 400 new models by 2023.

All the new models coming out will give Tesla a run for its money, predicts Shahar Bin-Nun, CEO of Tactile Mobility, an AV sensor tech company. Bin-Nun says he expected Tesla to still dominate the U.S. EV market in 2023, but that Ford, Hyundai and Kia will follow closely behind as they ramp up their lineups and production capacities.

We can also expect the market for secondhand EVs to creep up in 2023, which will make it much easier for people who are filthy rich to afford a zero-emission vehicle.

The software-defined vehicle will really take hold

Every automaker has been talking about the “software-defined vehicle” throughout 2022 as a concept that’s inherently linked to the electric vehicle. In 2023, we’ll really get a chance to see what that means.

General Motors, for example, will launch Ultifi early next year, its end-to-end vehicle software platform that promises OTA software updates, cloud connectivity and vehicle-to-everything communication. Ultifi will be the place where drivers can purchase apps, services and features – it’s an example of how automakers are increasingly trying to personalize vehicles to the individual’s needs.

This personalization will likely lead to an increase in subscription-based services in the car, says Will White, co-founder of Mapbox, a provider of online maps.

“We’ll also continue to see high demand for convenience-based services like in-car payments, where consumers will have a credit card on file in their app that pays for everything automotive-related,” said White.

On the backend, the software-defined vehicle will also dance with the metaverse. In 2022, a range of automakers, including Jaguar Land Rover, Nio, Polestar, Volvo and XPeng, announced plans to build software-defined vehicles on Nvidia’s Drive Orin system-on-a-chip. Automakers will in 2023 also rely on Nvidia’s recently upgraded its Omniverse platform, which stands to revolutionize everything from designing vehicles to the automotive product cycle. Using tech like this, automakers will increasingly build digital twins of both their vehicles and their production facilities in order to simulate anything from software upgrades within the vehicle to crash tests to factory efficiencies.

I guess we have to get used to saying Level 2+ ADAS

While we’re on the subject of software, automakers in 2023 will put much more investment into launching Level 2+ and Level 3 autonomous systems, which are basically really good advanced driver assistance systems. White says these systems will be a commonplace expectation in high-trim models.

Tesla will of course continue adding new features to its Autopilot and so-called “Full Self-Driving” softwares. But other automakers will come out with their own brands of impressive tech that will take care of more and more automated driving tasks.

Earlier this year, autonomous vehicle company Argo AI shutdown after Ford and Volkswagen pulled their investments. The IP was pretty much split between the two automakers, both of which said they were committed to pursuing near-term gains like L2+ and L3 systems. Rivian founder RJ Scaringe also said his company will focus on getting its own ADAS right.

Meanwhile in China, XPeng is rolling out the G9 SUV with its XNGP software, which the company describes as a “full scenario” ADAS that promises to automate highway driving, city driving and parking tasks.

More investment into getting charging right

J.D. Power analysts are expecting the market share of EVs in the U.S. to reach 12% next year, which is up from 7% today. If narrowing the scope to consumers that actually have access to EVs, that market share actually looks more like 20%.

Whatever the number, the fact remains that we’ll be seeing millions more EVs hit the streets in the U.S. next year. That means all of the ancillary services needed to keep them running will need to step up.

In 2023, we can expect to see investment – from government, utility and private firms – into charging infrastructure, energy storage and energy transmission.

Ensuring the EV transition is a smooth one isn’t just about building more EV chargers, although we grant, that’s a really important piece. Maintaining chargers will also be prioritized next year. A separate J.D. Power study earlier this year found that not only is availability of public charging still an obstacle, but often when you do find a charger, it’s broken. We predict there’ll be some tech, either from upstarts or existing EV charge players, that helps manage maintenance, servicing and upgrades for chargers.

In that same vein, all throughout 2022, every few months we stumble across some startup or utility company crying out that the electrical grid will never be able to handle all of the electric vehicles we’ll see in 2023. They’re probably right. So alongside energy management infrastructure, we expect to see more vehicle-to-grid software.

There were a few pilots in 2022, many of which were focused on V2G technology at home. Ford’s F-150 Lightning pickup truck is among a few vehicles that have promised to be able to power your home in the event of an outage. But we think as more fleets go electric, we’ll start to see those pilots happening in commercial settings at a wider scale.

The rise of EV fleets

We already saw many fleet operators begin to adopt EVs in 2022, as they aim to reach whatever carbon emissions goals they’ve set for themselves. Hertz, for example, plans to buy 65,000 Polestar vehicles, 100,000 Teslas and 175,000 General Motors vehicles over the next couple years to reach its goal of having 25% of its fleet electric by the end of 2024.

In 2023, those purchases will only ramp up, particularly as commercial EV makers get their production lines up and running.

GM’s BrightDrop, for example, has recently launched its CAMI Assembly plant in Ontario, which is expected to produce 50,000 of its Zevo delivery vans by 2025. BrightDrop has already secured over 25,000 reservations from customers like DHL and FedEx that are working towards net-zero goals.

Another commercial EV company Canoo plans to buy a vehicle manufacturing facility in Oklahoma City in order to ramp production of its Lifestyle Delivery Vehicle and bring those EVs to market next year for committed customers like NASA and Walmart.

An EV-plosion awaits in 2023, and it’ll be packed with tech by Rebecca Bellan originally published on TechCrunch

Research firm Fairwork India blasted Ola, Uber, Dunzo, PharmEasy and Amazon Flex in a report Tuesday, saying the firms scored zero in its assessment of whether they created fair conditions for their gig workers.

The research project, which collaborated with partners at the University of Oxford, said the aforementioned firms did not provide fair pay, fair contracts, fair management, fair representation or fair working conditions to their gig workers.

The firm studied 12 firms and granted unicorn Urban Company a score of seven out of 10, six to online grocer Bigbasket, five each to Flipkart and Swiggy, four to Zomato, two to grocery delivery firm Zepto and one to Tiger Global-backed delivery firm Porter.

“This year, only Bigbasket, Flipkart and Urban Company were awarded the first point because of the public commitments they have made to paying workers at least the hourly local minimum wage after factoring in work-related costs,” Fairwork India said in its fourth annual report.

“Bigbasket and Urban Company have operationalised this by committing to reimburse the difference between worker’s earnings per hour and the hourly local minimum wage after costs. Flipkart and Urban Company have committed to basing their pricing structure for workers on the hourly local minimum wage after costs. Flipkart has also undertaken steps to hold its third party service providers to the same commitment,” the report added.

Gig economy workers, whose participation to the workforce has significantly increased in recent years, aren’t extended the vast amount of employee benefits such as health insurance. Many researchers say the firms taking service from these workers are exploiting them and limiting corporate liabilities.

“The promise of flexibility of the digital platform economy raises as many questions about livelihoods as it offers opportunities. We hope the Fairwork report provides the basis for an interpretation of flexibility that allows for not merely the adaptability that platforms seek, but also the income and social security that workers lack,” said Professors Balaji Parthasarathy and Janaki Srinivasan, the principal investigators of the team, in a statement.

You can read the full-report here (PDF).

Uber and Amazon blasted for poor working conditions for gig workers in India by Manish Singh originally published on TechCrunch

Early indications show funding to digital health startups in Q4 2022 fell so much, they’re close to levels last seen in 2019.

But the dollar amounts don’t tell the whole story. How you grow as a digital healthcare company is just as important as if you grow at all.

A company built for the long term should have clinical experts as part of its leadership to ensure that care is always based on the patient’s medical needs as well as maintain quality control.

Here’s a framework that digital health startups can consider:

Bring clinicians into senior leadership

The best-case scenario for a digital health startup is to bring on a clinician as a co-founder.

I speak from experience. My co-founder is a triple-board-certified psychiatrist who brings clinical expertise to everything she does. From evaluating product roadmap decisions with our technology department to strategy discussions at board meetings and managing our entire clinical team, her contributions are vital to the health and direction of the company.

Outside the C-suite, hiring clinicians as senior leaders with responsibilities beyond clinical practice is invaluable. The key is to ensure clinicians know they will report to another clinician, not a non-clinical executive.

Non-clinical leaders, including founders and non-clinical C-suite executives, should practice what they preach. They should consistently loop in their clinical partners for business discussions even if they don’t have an obvious clinical impact.

The main benefits of taking this approach include:

The clinical and non-clinical partnership is more active from the jump;
Other team members and clinical staff will see and respect the inclusion;
Clinicians may uncover something that has an indirect but important clinical impact.

Beyond hiring clinicians in-house, startups should consider inviting clinicians to join their board of directors. Their presence on the board helps guide a company towards becoming an ethical and sustainable medical practice focused on helping patients rather than a technology company operating at the expense of patients.

This dedication to patient outcomes is a differentiator and should be reflected at every working level of a digital health startup.

Celebrate providers’ dedication

Dedicating resources and space to full-time providers allows them to focus more on patient care — the reason they got into medicine.

Digital health startups can incorporate clinical expertise into business models – here’s how by Ram Iyer originally published on TechCrunch

There are plenty of time-tracking apps for Mac that automatically log the hours you’ve spent signed-in. Some even offer granular data, telling you how much time you spent on a particular app. A new app called Balance is taking a slightly different approach to timekeeping, allowing users to manually punch in and punch out the time they are spending in front of a screen.

Balance hopes to help users build a set of healthy work habits rather than get granular data about their productivity. It won’t tell you long you had Slack, Microsoft Teams, Chrome or any other application open on your machine, but will offer general insights into your overall usage of the system and time spent in various sessions in a week.

To make this system work, Balance sends you a reminder if your machine has been on for more than five minutes but you have not clocked in. Clocking out is simple, too, just lock your Mac. Sadly, if your system goes to sleep, Balance doesn’t register a clock-out.

As there is no automatic tracking, the app can’t understand if you have taken a break even when you step away from the computer. So it will remind you to take a break after 60 minutes. You can easily fine-tune such settings as per your convenience.

Balance also offers you a Pomodoro timer (25 minutes on and 5 minutes off) through the Focus mode menu. The app lives in the menu bar of your Mac, so you can quickly access all the options. It shows the active time of the current session by default, but you can change it to the total session duration including breaks or time since the last break was taken.

Alexander Sandberg, the developer of Balance, says he built the app because he wanted a timekeeper that understands work-life balance. Working from home he often sat in front of his system way past his work hours, he told TechCrunch in an interview, and that’s when he thought of building Balance.

“I chose a manual clocking system for Balance because I believe it helps with creating a ‘ritual’ for checking in and out of work. Especially when working from home, it’s important to have something that helps you differentiate work time and non-work time. For instance, I’ve heard about people who go for a short walk to and from ‘the office’ at the beginning and at the end of the work days, even though their office is at home. This is to help the mind and body differentiate between life and work,” he told TechCrunch in an email.

While Balance is good for building the habit of clocking in and out, it could take a bit of time in getting used to. You might have many sessions that you forget to start or end. So you can end up with false positives on both ends.

Balance is available for free for everyone with the Pro version costing $2.49 a month (or $24.99 a year) as an introductory price. Paying customers will get features like session history with trends data. Balance also gives users an option to export their logs if they want to stop using the app or just want to analyze their data in a different way.

Sandberg said he’s building more pro features like a better session history overview with month and year; categorization and labeling of sessions; and app and website blocking to help users focus more.

Balance is a Mac timekeeper app that requires you to manually clock in your hours by Ivan Mehta originally published on TechCrunch

Baidu, the Chinese internet giant that became known for its search engines, is making some big strides in autonomous driving.

Starting this week, the public can ride its robotaxis in Wuhan between 7 am and 11 pm without safety drivers behind the wheel. Previously, its unmanned vehicles could only operate from 9 am to 5 pm in the city. The updated scheme is expected to cover one million customers in certain areas of Wuhan, a city of more than 10 million people.

Like most autonomous vehicle startups, Baidu combines a mix of third-party cameras, radars, and lidars to help its cars see better in low-visibility conditions, in contrast to Tesla’s vision-based solution.

In August, Baidu started offering fully driverless robotaxi rides, charging passengers at taxi rates. In Q3, Apollo Go, the firm’s robotaxi-hailing app, completed more than 474,000 rides, up 311% year over year. Accumulatively, Apollo Go had exceeded 1.4 million orders as of Q3.

That sounds like a potentially substantial revenue stream for Baidu, but one should take such figures with a grain of salt and ask: how many of these trips are subsidized by discounts? How many of them are repeatable, daily routes rather than one-off novelty rides taken by early adopters? To juice up performance numbers, it’s not uncommon to see Chinese robotaxi operators recruiting the public to ride in their vehicles.

It’s also tricky to tell which of China’s robotaxi upstarts have a lead at this stage. Their expansion isdependent on their relationship with the local city where they operate, and major cities often have the power to pass certain local legislations.

As one of the few remaining consumer internet sectors still with big room to grow, autonomous driving is getting warm support from local authorities nationwide. Case in point, Wuhan, an industrial hub in central China, is one of the first cities in the country to let robotaxis chauffeur the public without in-car safety operators. And now, the city seems to be comfortable with driverless cars roaming about even in low-light nighttime.

Setting aside a reasonable dose of skepticism, Baidu has indeed put a lot of effort into making the self-driving future arrive earlier. One of the moats it’s building is its visual-language model for identifying unseen or rare objects in long-tail scenarios. The AI is backed by Wenxin, the same large model that undergirds its text-to-image art platform.

“The model will enable autonomous vehicles to quickly make sense of an unseen object, such as special vehicle (fire truck, ambulance) recognition, plastic bag misdetection, and others,” Baidu previously explained. “In addition, Baidu’s autonomous driving perception model—a sub-model of the WenXin Big Model—leveraging more than 1 billion parameters, is able to dramatically improve the generalization potential of autonomous driving perception.”

Baidu starts offering nighttime driverless taxis in China by Rita Liao originally published on TechCrunch

Jakarta-based fintech Akulaku has raised $200 million from Mitsubishi UFJ Financial Group (MUFG), the largest bank in Japan. This is part of a strategic investment, with startup and MUFG planning to expand into new markets and product together in 2023. Earlier this year, Akulaku raised $100 million in funding from Siam Commercial Bank as part of another strategic investment. Its other backers include Ant Group (Akulaku launched a BNPL partnership earlier this year with Alipay+).

Akulaku, which operates in the Philippines and Malaysia in addition to Indoensia, offers a virtual credit card and installment shopping platform, as well as an investment platform and neobank. Founded in 2016, its target is to serve 50 million users by 2025.

As part of MUFG’s strategic investment, Akulaku has agreed to work with MUFG companies in Southeast on tech, product development, financing and distribution. MUFG is focused on growing its presence in the region, and earlier this year it purchased the Philippines and Indonesian units of Home Credit BV for 596 million euros. Its focus on Southeast Asia comes as homegrown banks, like Singapore’s DBS Group Holdings and Indonesia’s Bank Central Asia gain on MUFG in market cap.

In a statement, Kenichi Yamato, the managing executive officer and chief executive of MUFG Bank’s Global Commercial Banking Business Unit, said “Southeast Asia is key and a second market to MUFG. Our investment in Akulaku will further solidify our commitment in this region to meet growing financial needs of underserved customers.”

Jakarta-based fintech Akulaku raises $200M from Japan’s largest bank by Catherine Shu originally published on TechCrunch

High-growth companies often set significant goals, knowing full well that the idea of “overnight success” is for the storybooks. However, there is no better time than the middle of a market downturn to start planning for the leap from a private to a public company.

De-risking the path to going public requires strategic planning, which takes time. Companies with goals to go public in less than three years must therefore plan for it now — despite the downturn — to get the running start they’ll need to navigate the open market.

Let’s explore why this adverse economy is ideal for planning an IPO and what to do about it.

Growth investors have recently pulled back

Carta reports that private fundraising levels have declined across the U.S. from a record-breaking 2021. Unsurprisingly, late-stage companies have experienced the brunt of this blow.

Market experts are currently encouraging leaders not to pin their hopes on venture capital dry powder, even though there’s plenty of it. As the graph below indicates, the size of late-stage funding rounds has shrunk.

Although few enjoy market downturns, how this one unfolds can deliver insights to late-stage companies that pay attention. On one hand, many leaders are embracing the message of the Sequoia memo. We can agree with their ideas to prioritize profits over growth — scaling is different from what it used to be, and we must swallow that jagged pill.

On the other hand, cost-cutting and giving up hope of fundraising isn’t all doom and gloom. After all, when there is money to be found, some innovative founder will find it. We see it every day; only now, the path looks different.

Market downturns spur valuation corrections

Course-correcting is a concept frequently discussed amid market downturns. The pendulum swings one way for a period, then begins its journey toward a more balanced standard. In this case, the open market thrived on bloated valuations — most startups were overvalued before 2021.

Furthermore, many stated that 2021 was a miracle year, especially as VC investment nearly doubled to $643 billion. The U.S. sprouted more than 580 new unicorns and saw over 1,030 IPOs (over half were SPACs), significantly higher than the year before. This year has only welcomed about 170 public listings.

High-growth startups should start de-risking their path to IPO now by Ram Iyer originally published on TechCrunch

As a rather commercially successful author once wrote, “the night is dark and full of terrors, the day bright and beautiful and full of hope.” It’s fitting imagery for AI, which like all tech has its upsides and downsides.

Art-generating models like Stable Diffusion, for instance, have led to incredible outpourings of creativity, powering apps and even entirely new business models. On the other hand, its open source nature lets bad actors to use it to create deepfakes at scale — all while artists protest that it’s profiting off of their work.

What’s on deck for AI in 2023? Will regulation rein in the worst of what AI brings, or are the floodgates open? Will powerful, transformative new forms of AI emerge, a la ChatGPT, disrupt industries once thought safe from automation?

Expect more (problematic) art-generating AI apps

With the success of Lensa, the AI-powered selfie app from Prisma Labs that went viral, you can expect a lot of me-too apps along these lines. And expect them to also be capable of being tricked into creating NSFW images, and to disproportionately sexualize and alter the appearance of women.

Maximilian Gahntz, a senior policy researcher at the Mozilla Foundation, said he expected integration of generative AI into consumer tech will amplify the effects of such systems, both the good and the bad.

Stable Diffusion, for example, was fed billions of images from the internet until it “learned” to associate certain words and concepts with certain imagery. Text-generating models have routinely been easily tricked into espousing offensive views or producing misleading content.

Mike Cook, a member of the Knives and Paintbrushes open research group, agrees with Gahntz that generative AI will continue to prove a major — and problematic — force for change. But he thinks that 2023 has to be the year that generative AI “finally puts its money where its mouth is.”

“It’s not enough to motivate a community of specialists [to create new tech] — for technology to become a long-term part of our lives, it has to either make someone a lot of money, or have a meaningful impact on the daily lives of the general public,” Cook said. “So I predict we’ll see a serious push to make generative AI actually achieve one of these two things, with mixed success.”

Artists lead the effort to opt out of data sets

DeviantArt released an AI art generator built on Stable Diffusion and fine-tuned on artwork from the DeviantArt community. The art generator was met with loud disapproval from DeviantArt’s longtime denizens, who criticized the platform’s lack of transparency in using their uploaded art to train the system.

The creators of the most popular systems — OpenAI and Stability AI — say that they’ve taken steps to limit the amount of harmful content their systems produce. But judging by many of the generations on social media, it’s clear that there’s work to be done.

“The data sets require active curation to address these problems and should be subjected to significant scrutiny, including from communities that tend to get the short end of the stick,” Gahntz said, comparing the process to ongoing controversies over content moderation in social media.

Stability AI, which is largely funding the development of Stable Diffusion, recently bowed to public pressure, signaling that it would allow artists to opt out of the data set used to train the next-generation Stable Diffusion model. Through the website HaveIBeenTrained.com, rightsholders will be able to request opt-outs before training begins in a few weeks’ time.

OpenAI offers no such opt-out mechanism, instead preferring to partner with organizations like Shutterstock to license portions of their image galleries. But given the legal and sheer publicity headwinds it faces alongside Stability AI, it’s likely only a matter of time before it follows suit.

The courts may ultimately force its hand. In the U.S. Microsoft, GitHub and OpenAI are being sued in a class action lawsuit that accuses them of violating copyright law by letting Copilot, GitHub’s service that intelligently suggests lines of code, regurgitate sections of licensed code without providing credit.

Perhaps anticipating the legal challenge, GitHub recently added settings to prevent public code from showing up in Copilot’s suggestions and plans to introduce a feature that will reference the source of code suggestions. But they’re imperfect measures. In at least one instance, the filter setting caused Copilot to emit large chunks of copyrighted code including all attribution and license text.

Expect to see criticism ramp up in the coming year, particularly as the U.K. mulls over rules that would that would remove the requirement that systems trained through public data be used strictly non-commercially.

Open source and decentralized efforts will continue to grow

2022 saw a handful of AI companies dominate the stage, primarily OpenAI and Stability AI. But the pendulum may swing back towards open source in 2023 as the ability to build new systems moves beyond “resource-rich and powerful AI labs,” as Gahntz put it.

A community approach may lead to more scrutiny of systems as they are being built and deployed, he said: “If models are open and if data sets are open, that’ll enable much more of the critical research that has pointed to a lot of the flaws and harms linked to generative AI and that’s often been far too difficult to conduct.”

Examples of such community-focused efforts include large language models from EleutherAI and BigScience, an effort backed by AI startup Hugging Face. Stability AI is funding a number of communities itself, like the music-generation-focused Harmonai and OpenBioML, a loose collection of biotech experiments.

Money and expertise are still required to train and run sophisticated AI models, but decentralized computing may challenge traditional data centers as open source efforts mature.

BigScience took a step toward enabling decentralized development with the recent release of the open source Petals project. Petals lets people contribute their compute power, similar to Folding@home, to run large AI language models that would normally require an high-end GPU or server.

“Modern generative models are computationally expensive to train and run. Some back-of-the-envelope estimates put daily ChatGPT expenditure to around $3 million,” Chandra Bhagavatula, a senior research scientist at the Allen Institute for AI, said via email. “To make this commercially viable and accessible more widely, it will be important to address this.”

Chandra points out, however, that that large labs will continue to have competitive advantages as long as the methods and data remain proprietary. In a recent example, OpenAI released Point-E, a model that can generate 3D objects given a text prompt. But while OpenAI open sourced the model, it didn’t disclose the sources of Point-E’s training data or release that data.

“I do think the open source efforts and decentralization efforts are absolutely worthwhile and are to the benefit of a larger number of researchers, practitioners and users,” Chandra said. “However, despite being open-sourced, the best models are still inaccessible to a large number of researchers and practitioners due to their resource constraints.”

AI companies buckle down for incoming regulations

Regulation like the EU’s AI Act may change how companies develop and deploy AI systems moving forward. So could more local efforts like New York City’s AI hiring statute, which requires that AI and algorithm-based tech for recruiting, hiring or promotion be audited for bias before being used.

Chandra sees these regulations as necessary especially in light of generative AI’s increasingly apparent technical flaws, like its tendency to spout factually wrong info.

“This makes generative AI difficult to apply for many areas where mistakes can have very high costs — e.g. healthcare. In addition, the ease of generating incorrect information creates challenges surrounding misinformation and disinformation,” she said. “[And yet] AI systems are already making decisions loaded with moral and ethical implications.”

Next year will only bring the threat of regulation, though — expect much more quibbling over rules and court cases before anyone gets fined or charged. But companies may still jockey for position in the most advantageous categories of upcoming laws, like the AI Act’s risk categories.

The rule as currently written divides AI systems into one of four risk categories, each with varying requirements and levels of scrutiny. Systems in the highest risk category, “high-risk” AI (e.g. credit scoring algorithms, robotic surgery apps), have to meet certain legal, ethical and technical standards before they’re allowed to enter the European market. The lowest risk category, “minimal or no risk” AI (e.g. spam filters, AI-enabled video games), imposes only transparency obligations like making users aware that they’re interacting with an AI system.

Os Keyes, a Ph.D. Candidate at the University of Washington, expressed worry that companies will aim for the lowest risk level in order to minimize their own responsibilities and visibility to regulators.

“That concern aside, [the AI Act] really the most positive thing I see on the table,” they said. “I haven’t seen much of anything out of Congress.”

But investments aren’t a sure thing

Gahntz argues that, even if an AI system works well enough for most people but is deeply harmful to some, there’s “still a lot of homework left” before a company should make it widely available. “There’s also a business case for all this. If your model generates a lot of messed up stuff, consumers aren’t going to like it,” he added. “But obviously this is also about fairness.”

It’s unclear whether companies will be persuaded by that argument going into next year, particularly as investors seem eager to put their money beyond any promising generative AI.

In the midst of the Stable Diffusion controversies, Stability AI raised $101 million at an over-$1 billion valuation from prominent backers including Coatue and Lightspeed Venture Partners. OpenAI is said to be valued at $20 billion as it enters advanced talks to raise more funding from Microsoft. (Microsoft previously invested $1 billion in OpenAI in 2019.)

Of course, those could be exceptions to the rule.

Outside of self-driving companies Cruise, Wayve and WeRide and robotics firm MegaRobo, the top-performing AI firms in terms of money raised this year were software-based, according to Crunchbase. Contentsquare, which sells a service that provides AI-driven recommendations for web content, closed a $600 million round in July. Uniphore, which sells software for “conversational analytics” (think call center metrics) and conversational assistants, landed $400 million in February. Meanwhile, Highspot, whose AI-powered platform provides sales reps and marketers with real-time and data-driven recommendations, nabbed $248 million in January.

Investors may well chase safer bets like automating analysis of customer complaints or generating sales leads, even if these aren’t as “sexy” as generative AI. That’s not to suggest there won’t be big attention-grabbing investments, but they’ll be reserved for players with clout.

What to expect from AI in 2023 by Kyle Wiggers originally published on TechCrunch