Connect with us


Vera wants to use AI to cull generative models’ worst behaviors



Liz O’Sullivan is on a mission to make AI “a little bit safer,” in her own words.

A member of the National AI Advisory Committee, which drafts recommendations to the White House and Congress on how to foster AI adoption while regulating its risks, O’Sullivan spent 12 years on the business side of AI startups overseeing data labeling and operations and customer success. In 2019, she took a job at the Surveillance Technology Oversight Project, mounting campaigns to protect New Yorkers’ civil liberties, and co-founded Arthur AI, a startup that partners with civil society and academy to shine light into AI’s “black box.”

Now, O’Sullivan is gearing up for her next act with Vera, a startup building a toolkit that allows companies to establish “acceptable use policies” for generative AI — the type of AI models that generate text, images, music and more — and enforce these policies across open source and custom models.

Vera today closed a $2.7 million funding round led by Differential Venture Partners with participation from Essence VC, Everywhere VC, Betaworks, Greycroft and ATP Ventures. Bringing Vera’s total raised to $3.3 million, the new cash will be put toward growing Vera’s five-person team, R&D and scaling enterprise deployments, O’Sullivan says.

“Vera was founded because we’ve seen, firsthand, the power of AI to address real problems, just as we’ve seen the wild and wacky ways it can cause damage to companies, the public and the world,” O’Sullivan told TechCrunch in an email interview. “We need to responsibly shepherd this technology into the world, and as companies race to define their generative AI strategies, we’re entering an age where it’s critical that we move beyond AI principles and into practice. Vera is a team that can actually help.

O’Sullivan co-founded Vera in 2021 with Justin Norman, formerly a research scientist at Cisco, a lead data scientist in Cloudera’s AI research lab and the VP of data science at Yelp. In September, Norman was appointed a member of the Department of the Navy Science and Technology board, which provides advice and counsel to the U.S. Navy on matters and policies relating to scientific, technical and related functions,

Vera’s platform attempts to identify risks in model inputs — for example, a prompt like “write a cover letter for a software engineering role” to a text-generating model — and block, redact or otherwise transform requests that might contain things like personally identifiable information, security credentials, intellectual property and prompt injection attacks. (Prompt injection attacks, essentially carefully-worded malicious prompts, are often used to “trick” models into bypassing safety filters.)

Vera also places constraints on what models can “say” in response to prompts, according to O’Sullivan, giving companies greater control over the behavior of their models in production.

How does Vera achieve this? By using what O’Sullivan describes as “proprietary language and vision models” that sit between users and internal or third-party models (e.g. OpenAI’s GPT-4) and detect problematic content. Vera can block “inappropriate” prompts to — or answers from a model in any form, O’Sullivan claims, whether text, code, image or video.

“Our deep tech approach to enforcing policies goes beyond passive forms of documentation and checklists to address the direct points at which these risks occur,” O’Sullivan said. “Our solution … prevents riskier responses that may include criminal material or encourage users to self-harm.”

Companies are certainly encountering challenges — mainly compliance-related — in adopting generative AI models for their purposes. They’re worried about their confidential data ending up with developers who trained the models on user data, for instance; in recent months, major corporations including Apple, Walmart and Verizon have banned employees from using tools like OpenAI’s ChatGPT.

And offensive models are obviously bad for publicity. No brand wants the text-generating model powering their customer service chatbot, say, to spout racial epithets or give self-destructive advice.

But this reporter wonders if Vera’s approach is as reliable as O’Sullivan suggests.

No model is perfect — not even Vera’s — and it’s been demonstrated time and time again that content moderation models are prone to a whole host of biases. Some AI models trained to detect toxicity in text see phrases in African-American Vernacular English, the informal grammar used by some Black Americans, as disproportionately “toxic.” Meanwhile, certain computer vision algorithms have been found to label thermometers held by Black people as “guns” while labeling thermometers held by light-skinned subjects as “electronic devices.”

To be fair to O’Sullivan, she doesn’t claim Vera’s models are bulletproof — only that they can cull the worst of a generative AI models’ behaviors. There may be some truth to that (depending on the model, at least) — and the degree to which Vera has iterated and refined its own models.

“Today’s AI hype cycle obscures the very serious, very present risks that affect humans alive today,” O’Sullivan said. “Where AI overpromises, we see real people hurt by unpredictable, harmful, toxic and potentially criminal model behavior … AI is a powerful tool and like any powerful tool, should be actively controlled so that its benefits outweigh these risks, which is why Vera exists.”

Vera’s possible shortcomings aside, the company has competition in the nascent market for model-moderating tech.

Similar to Vera, Nvidia’s NeMo Guardrails and Salesforce’s Einstein Trust Layer attempt to prevent text-generating models from retaining or regurgitating sensitive data, such as customer purchase orders and phone numbers. Microsoft provides an AI service to moderate text and image content, including from models. Elsewhere, startups like HiddenLayer, DynamoFL and Protect AI are creating tooling to defend generative AI models against prompt engineering attacks.

So far as I can tell, Vera’s value proposition is that it tackles a whole range of generative AI threats at once — or promises to at the very least. Assuming that the tech works as advertised, that’s bound to be attractive for companies in search of a one-stop content moderation, AI-model-attack-fighting shop.

Indeed, O’Sullivan says that Vera already has a handful of customers. The waitlist for more opens today.

“CTOs, CISOs and CIOs all over the world are struggling to strike the ideal balance between AI-enhanced productivity and the risks these models present,” O’Sullivan said. “Vera unlocks generative AI capabilities with policy enforcement that can be transferred not just to today’s models, but to future models without the vendor lock-in that occurs when you choose a one-model or one-size-fits-all approach to generative AI.”

Source link

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *


SpaceX Rocket Suffers Engine Failure In Starlink Mission, Elon Musk Shares What’ll Happen Next




SpaceX’s Falcon 9 rocket suffered a rare failure on Friday. The launcher lifted off with 20 Starlink satellites at 8:06 am IST from California but failed to deploy them in the intended orbit.

According to SpaceX, the second stage of Falcon 9 did not complete its second burn necessary for reaching the desired orbit. The booster did its job and safely landed on the droneship in the Pacific Ocean. It was SpaceX‘s 70th mission this year.

Currently, the satellites are in a lower orbit than planned.

“SpaceX has made contact with 5 of the satellites so far and is attempting to have them raise orbit using their ion thrusters,” the company said in a statement on X.

ALSO SEE: Japan’s First Private Rocket Launch Ends In Massive Explosion, Video Goes Viral

SpaceX CEO Elon Musk said that it might not work “but it’s worth the shot. He also said that the satellites might fall toward Earth and burn up in the atmosphere if their thrusters are overpowered by the atmospheric drag.

In another post, Musk revealed that the upper stage engine exploded for unknown reasons while trying to raise the orbit. “Team is reviewing data tonight to understand root cause,” he said.

Notably, the reputed rocket by SpaceX has suffered only one full in-flight failure during a mission to the International Space Station in 2015.

ALSO SEE: Chinese Rocket Launches Accidentally, Crashes With Massive Explosion Seconds Later

(Image: SpaceX)

Source link

Continue Reading


China Plans To Destroy An Asteroid For Planetary Defense Mission By 2030: Report




After the impeccable success of NASA’s Double Asteroid Redirection Test (DART) mission, China is now planning to deflect an asteroid later this decade. According to The Planetary Society, the China National Space Administration (CNSA) is preparing its first planetary defense test which is expected to launch by 2030.

Scientists have proposed the near-Earth asteroid 2015 XF261 as a candidate for the mission which will include two probes. One will ram into the space rock to deflect it and the second will conduct impact assessment.

The asteroid 2015 XF261 measures about 100 feet or 30 metres in diameter and it made a close flyby of our planet earlier this month. According to NASA’s Jet Propulsion Laboratory (JPL), the asteroid was about 50 million kilometres from our planet on July 9 and was travelling at a speed of around 42,000 kilometres per hour.

ALSO SEE: NASA Drills Freaky Scenario Where Elusive Asteroid Heads Towards Earth

Apart from China, Japan is also eyeing a ‘kinetic impact’ test mission to deflect an asteroid. The Japan Aerospace Exploration Agency (JAXA) reportedly has plans to repurpose its Hayabusa2 spacecraft to collide with 1998 KY26. The probe which launched in 2014 is expected to rendezvous with the space rock in 2031 and potentially change its orbit.

The said missions by the two nations are driven by the success of NASA’s DART mission launched in 2021 which proved that smacking an asteroid can deflect them. It collided with Dimorphos which circles a larger rock Didymos in September 2022 and changed its orbit by about 32 minutes.

These missions are of immense importance as they enable technologies that could save Earth from a planet-killing asteroid. While predicting an impending asteroid armageddon is predictable, it is also the gravest threat that humanity faces.

ALSO SEE: Collision Of NASA’s DART With Asteroid Dimorphos Changed Its Shape; Finding Excites Scientists

Source link

Continue Reading


Uzbekistan mobile bank TBC raises $38.2M to expand its financial products




Uzbekistan’s only mobile-exclusive bank, TBC Bank Uzbekistan, owned by London Stock Exchange-listed TBC Bank Group, has raised $38.2 million in a fresh equity investment. It plans to expand its local presence in the country and introduce new financial products as well.

TBC Bank Group has led the latest funding in TBC Bank Uzbekistan by infusing $23 million, while shareholders European Bank for Reconstruction and Development (EBRD) and World Bank’s International Finance Corporation (IFC) have participated in the round by investing $7.6 million each.

After serving customers in Georgia, TBC Bank Group decided to expand outside that country’s borders in 2019 and found Uzbekistan as its first international market. The bank started its Uzbekistan operations in 2020 through a separate entity, TBC Group Uzbekistan, which launched the mobile-only bank after its debut, with no physical branches in the country. The plan was to leverage Uzbekistan’s increasing digitization effort and foresee business growth in the country, which has the biggest population after Russia and Ukraine in the region — the second-largest among all the Commonwealth of Independent States countries — and has upright economic and socio-demographics.

“Before TBC came in, there were no banking apps in Uzbekistan … Fast-forward four years, most of the banks have got a mobile app, but TBC is far ahead of the field,” said Oliver Hughes, head of international business at TBC Group, in an exclusive interview.

According to official data, Uzbekistan has a 70% smartphone penetration rate and a 77% internet penetration rate; 59% of its population of 37 million is under 30 years old, making it a viable market for a mobile-specific business.

TBC Bank Uzbekistan offers a mobile app through which customers can open bank accounts and access services, including cash loans and deposits. This omits the requirement of physically going to a bank branch to access banking.

Hughes told TechCrunch that a couple of years ago, customers in Uzbekistan typically had to visit their bank and stand in a queue to get any of their banking work done.

Alongside the mobile-only bank, TBC Group Uzbekistan owns Payme, the digital payments app for individual users and small businesses, as well as the Sharia-compliant credit business called Payme Nasiya. To broaden its coverage, it looks to integrate some experiences from these two businesses within the bank or sync them with the bank’s operations.

For instance, through its app, TBC Group Uzbekistan will offer tips, recommendations and user-generated content on local events, entertainment, concerts and travel to provide complementary services that are not strictly financially related. Some of these features will first arrive on the Payme app but will be available to the TBC Bank Uzbekistan customers over time.

Similarly, Payme Nasiya currently serves Uzbek customers with its point-of-sale and installment loans. To expand the credit business, it will introduce e-commerce and offline buy now, pay later. This is expected to attract more local businesses and eventually help the mobile bank gain more customers.

In addition to the new financial products in the pipeline, TBC Group Uzbekistan plans to bring AI experiences to its mobile bank. Hughes told TechCrunch the group has built a large language model predominantly using its customer dataset and is working on a voice assistant to deliver banking and financial services through a chatbot integrated within its app.

In the fall, TBC Bank Uzbekistan will use the fresh funding to add credit cards and an insurance product next year, Hughes said.

The bank’s roadmap includes additional services such as current accounts, as well as accounting, offline payments, e-commerce payments and lending specifically for small and medium enterprise customers, Hughes added.

“This investment will allow us to further capitalize on the immense opportunities in Uzbekistan, a fast-growing country with a population of over 37 million people where TBC UZ continues to leverage its growth momentum,” said Vakhtang Butskhrikidze, CEO of TBC Bank Group, in a prepared statement.

At the end of 2022, TBC Bank Uzbekistan broke even, and 2023 was the bank’s first full year of profit. As of March 2024, the bank had a user base of 4.8 million unique registered users. It also recorded monthly active users of 1.2 million in the first quarter of 2024.

Overall, TBC Group Uzbekistan, with a registered user base of 15 million users, reached profitability two years after launch and recorded 85% year-on-year revenue growth in the first quarter of this year. The company achieved gross loans of $296 million and deposits of $216 million through all three of its subsidiaries. Its net profit hit $23 million for the financial year 2023, most of which came from Payme. However, TBC did not disclose the mobile bank’s revenues or profits.

“TBC UZ’s impressive growth trajectory and innovative approach align with our mission to support sustainable economic development in the region,” said Andi Aranitasi, head of Uzbekistan, EBRD.

Hughes said that by the end of 2025, TBC Group Uzbekistan is projected to generate $75 million in net profit, most of which would come from TBC Bank Uzbekistan.

“We are encouraged with the progress TBC UZ has made so far and remain confident in its potential to contribute to economic growth and financial inclusion in Uzbekistan,” said Neil McKain, country manager, Uzbekistan, IFC.

Source link

Continue Reading


Copyright © 2023 Dailycrunch. & Managed by Shade Marketing & PR Agency