Category Archives: Ai

AI for AP: High school students create tool that acts as tech tutor for advanced placement courses – GeekWire

Professor AI co-founders Pradyu Kandala, left, and Abhay Chebium. (Professor AI Photos)

Frustrated by inadequate resources to help students study for advanced placement courses, two Seattle-area high school teens turned to technology, and put AI to work on AP.

Pradyu Kandala and Abhay Chebium are juniors at Eastlake High School in Sammamish, Wash., and the co-founders of Professor AI, an ed-tech tool they started building last summer.

We started this because we experienced the problem, said Kandala, an aspiring entrepreneur who is the startups CEO. Chebium is the CTO. We know that if we can use this for our homework, everybodys going to want to use it for homework.

Professor AI is like a ChatGPT for AP courses, according to Kandala, and is designed to help students via study guides, practice tests or quizzes, course-specific chats, and more. Using AI models from OpenAI, Claude, Anthropic, and Mistral AI, Professor AI is trained on AP-specific curriculum such as AP rubrics, textbooks, and various niche questions.

As artificial intelligence technology has increasingly made its way into the classroom, concern has also increased over the use of ChatGPT and similar software that can automatically generate content and answer prompts. Seattle Public Schools originally banned ChatGPT on student devices before making it available to ensure students have access to this transformative technology.

Professor AI is not built to just give kids answers and help them cheat.

Kandala said the goal is to use tech to replace three obstacles impeding how students currently study for AP courses the college-level classes that some kids take in high school to prepare for higher education. Those obstacles include teachers with too many students and not enough time; tutors that are expensive and less accessible; and self study, which can be overwhelming for lots of kids.

With plenty of competition online promising all kinds of help with schoolwork, Kandala said he feels like he and Chebium are uniquely qualified to tackle the problem. They know what AP students need, they have access to potential customers on a daily basis at school, and they can iterate and make a better product faster because of their connections.

If I want to talk to an AP student, I can go to class tomorrow and talk to 30 of them instantly, Kandala said. Compared to a conglomerate, it would take them months to get in contact with people and have proper conversations.

Kandala and Chebium are already enrolled in a New York-based accelerator program with Antler.

The company has raised $20,000 so far and is actively looking for more investors.

The startups monetization plan is based on a subscription model. Theres a free plan, a $14/month scholar plan and an elite plan for $29/month that promises unlimited access to chats, messages, classes, math problems and more. In a nod to their student customer base, Professor AI notes that the scholar plan would cost one Chipotle.

Professor AI has a small number of mostly high schoolers working as developers and in marketing/content.

Read more from the original source:

AI for AP: High school students create tool that acts as tech tutor for advanced placement courses - GeekWire

Apple reportedly courting OpenAI to develop AI features for iPhone – Cointelegraph

Apple is reportedly on the hunt for a big tech partner to help shore up its flagging artificial intelligence (AI) offerings, with a specific focus on bringing th iPhones capabilities up to speed.

Per a report from Bloomberg citing multiple sources, Apple and OpenAI are deep into discussions concerning the further integration of generative AI technologies on iOS.

However, previous reports also indicate that Apple is still shopping around with Google on the hook as another potential cloud partner.

The Cupertino companys AI ambitions appear to have fizzled over the past few years. Where once Apples walled garden approach to AI development seemed to indicate that it might have some mysterious tricks up its sleeve now, in 2024, the companys flagship offering, the iPhone, is starting to show signs of lagging behind its AI-first competitors.

Arguably, the shift from internal development to external partnerships was signalled in May of 2022 when Ian Goodfellow, often referred to as the GANfather for his contributions to the field of AI, left his post as director of machine learning over Apples return to office policy.

In the time since Goodfellows departure, Apple has remained busy in the AI sector. The company recently released a bevy of open-source large language models, eight in total, complete with code and training logs.

While this signals a continuing dedication to AI and machine learning, its unlikely that any of the released models could be considered competitive with available closed models such as OpenAIs ChatGPT and Googles Gemini.

As such, its possible that Apple is trying to finalize its generative AI partnerships ahead of its annual Worldwide Developers Conference (WWDC) in June.

The WWDC event is the typical showcase for the companys biggest and most impactful products and, so far, 2024 has been a relatively quiet year for Apple.

Meanwhile, the companys largest competitor, Microsoft, has had a metaphorical liscene to print money ever since it partnered up with AI sector leader OpenAI. As of current, Microsoft has a commanding lead as the worlds most valuable company by market capitalization with a total value of $3.019 trillion against Apples second place score of $2.614 trillion.

Related: Doctors turn to Apple Vision Pro headset to practice surgery amid cadaver shortage

View post:

Apple reportedly courting OpenAI to develop AI features for iPhone - Cointelegraph

Rabbit R1 hands-on: early tests with the $199 AI gadget – The Verge

There were times I wasnt sure the Rabbit R1 was even a real thing. The AI-powered, Teenage Engineering-designed device came out of nowhere to become one of the biggest stories at CES, promising a level of fun and whimsy that felt much better than some of the more self-serious AI companies out there. CEO Jesse Lyu practically promised the world in this $199 device.

Well, say this for Rabbit: its real. Last night, I went to the swanky TWA Hotel in New York City, along with a few hundred reporters, creators, and particularly enthusiastic R1 buyers. After a couple of hours of photo booths, specialty cocktails, and a rousing keynote and demo from Lyu in which he made near-constant reference to and fun of the Humane AI Pin we all got our R1s to take home. Ive been using mine ever since, and I have some thoughts. And some questions.

From a hardware perspective, the R1 screams kinda meh Android phone. Here are the salient specs: its about three inches tall and wide and a half-inch thick. It weighs 115 grams, which is about two-thirds as much as the iPhone 15. It has a 2.88-inch screen, runs on a 2.3GHz MediaTek MT6765 processor, and has 128 gigs of storage and four gigs of RAM. It has a speaker on the back, two mics on the top, and a SIM card slot on the side right next to the USB-C charging port.It only comes in one color, a hue Rabbit calls leuchtorange but is often known as brilliant orange or luminous orange. Its definitely orange, and its definitely luminous.

At this point, the best way I can describe the R1 is like a Picasso painting of a smartphone: it has most of the same parts, just laid out really differently. Instead of sitting on top or in the back, the R1s camera sits in a cutout space on the right side of the device, where it can spin its lens to face both toward and away from you.

The R1 is like a Picasso painting of a smartphone

After spending a few hours playing with the device, I have to say: its pretty nice. Not luxurious, or even particularly high-end, just silly and fun. Where Humanes AI Pin feels like a carefully sculpted metal gem, the R1 feels like an old-school MP3 player crossed with a fidget spinner. The wheel spins a little stiffly for my taste but smoothly enough, the screen is a little fuzzy but fine, and the main action button feels satisfying to thump on.

When I first got the device and connected it to Wi-Fi, it then immediately asked me to sign up for an account at Rabbithole, the R1s web portal. I did that, scanned a QR code with the R1 to get it synced up, and immediately did a software update. I spent that time logging in to the only four external services the R1 currently connects to: Spotify, Uber, DoorDash, and Midjourney.

Once I was eventually up and running, I started chatting with the R1. So far, it does a solid job with basic AI questions: it gave me lots of good information about this weeks NFL draft, found a few restaurants near me, and knew when Herbert Hoover was president. This is all fairly basic ChatGPT stuff, and theres some definite lag as it fetches answers, but I much prefer the interface to the Humane AI Pin because theres a screen, and you can see the thing working so the AI delays dont feel quite so interminable.

Because theres a screen, the AI delays dont feel quite so interminable

Almost immediately, though, I started running into stuff the R1 just cant do. It cant send emails or make spreadsheets, though Lyu has been demoing both for months. Rabbithole is woefully unfinished, too, to the point I was trying to tap around on my phone and it was instead moving a cursor around a half-second after every tap. Thats a good reminder that the whole thing is running on a virtual machine storing all your apps and credentials, which still gives me security-related pause.

Oh, and heres my favorite thing that has happened on the R1 so far: I got it connected to my Spotify account, which is a feature Im particularly excited about. I asked for Beyoncs new album, and the device excitedly went and found me Crazy in Love a lullaby version, from an artist called Rockabye Baby! So close and yet so far. It doesnt seem to be able to find my playlists, either, or skip tracks. When I said, Play The 1975, though, that worked fine and quickly. (The speaker, by the way, is very much crappy Android phone quality. Youre going to want to use that Bluetooth connection.)

The R1s Vision feature, which uses the camera to identify things in the scene around you, seems to work fine as long as all you want is a list of objects in the scene. The device cant take a photo or video and doesnt seem to be able to do much else with what it can see.

When youre not doing anything, the screen shows the time and that bouncing rabbit-head logo. When you press and hold the side button to issue a command, the time and battery fade away, and the rabbits ears perk up like its listening. Its very charming! The overall interface is simple and text-based, but its odd in spots: its not always obvious how to go back, for instance, and you only get to see a line or two of text at a time at the very bottom of the screen, even when theres a whole paragraph of answer to read.

Rabbits roadmap is ambitious: Lyu has spent the last few months talking about all the things the R1s so-called Large Action Model can do, including learning apps and using them for you. During last nights event, he talked about opening up the USB-C port on the device to allow accessories, keyboards, and more. Thats all coming eventually. Supposedly. For now, the R1s feature set is much more straightforward. You can use the device to play music, get answers to questions, translate speech, take notes, summon an Uber, and a few other things.

That means theres still an awful lot the R1 cant do and a lot I have left to test. (Anything you want to know about, by the way, let me know!) Im particularly curious about its battery life, its ability to work with a bad connection, whether it heats up over time, and how it handles more complex tasks than just looking up information and ordering chicken nuggets. But so far, this thing seems like its trying to be less like a smartphone killer and more like the beginnings of a useful companion. Thats probably as ambitious as it makes sense to be right now though Lyu and the Rabbit folks have a lot of big promises to eventually live up to and not a lot of time to do so.

Photography by David Pierce / The Verge

Link:

Rabbit R1 hands-on: early tests with the $199 AI gadget - The Verge

Baltimore coach allegedly used AI voice cloning to get principal fired – The Verge

A physical education teacher and former athletic director of a Baltimore County high school has been arrested and charged with using an AI voice cloning service to frame the schools principal.

The Baltimore Banner reports that Baltimore County police believe a recording that circulated on social media in January with purported audio of Pikesville High School principal Eric Eiswert making racist and antisemitic comments was fake. Experts told The Baltimore Banner and police that the recording, which briefly resulted in Eiswerts suspension, has a flat tone, unusually clean background sounds, and lack of consistent breathing sounds or pauses.

Baltimore County police traced the recording to Dazhon Darien, a former athletic director at the school whose name was also mentioned in the audio clip. He allegedly used school computers to access OpenAI tools and Microsoft Bing Chat services, as reported by WBAL-TV 11 and NBC News. He was also linked to the audios release via an email address and associated recovery phone number.

It is not clear what AI voice platform Darien allegedly used.

The police arrested Darien on Thursday at the airport and said in a statement, Its believed Mr. Darien, who was an Athletic Director at Pikesville High School, made the recording to retaliate against Mr. Eiswert who at the time was pursuing an investigation into the potential mishandling of school funds. He has been released after posting bail and faces charges including theft (for the issue with school funds), disturbing the operations of a school, retaliation against a witness, and stalking.

In this fraught environment, OpenAI decided in March to withhold its AI text-to-voice generation platform, Voice Engine, from public use. The service, which only requires a 15-minute audio clip to clone someones voice, is only available to a limited number of researchers due to the lack of guardrails around the technology.

US lawmakers have filed, but not yet passed, several bills like the No Fakes Act and the No AI Fraud Act that seek to prevent technology companies from using an individuals face, voice, or name without their permission.

Update, April 25th: Clarified that Pikesville High School is in the city of Pikesville in Baltimore County, Maryland, and added details about Darien and Thursdays arrest.

The rest is here:

Baltimore coach allegedly used AI voice cloning to get principal fired - The Verge

Racist AI Deepfake of Baltimore Principal Leads to Arrest – The New York Times

A high school athletic director in the Baltimore area was arrested on Thursday after he used artificial intelligence software, the police said, to manufacture a racist and antisemitic audio clip that impersonated the schools principal.

Dazhon Darien, the athletic director of Pikesville High School, fabricated the recording including a tirade about ungrateful Black kids who cant test their way out of a paper bag in an effort to smear the principal, Eric Eiswert, according to the Baltimore County Police Department.

The faked recording, which was posted on Instagram in mid-January, quickly spread, roiling Baltimore County Public Schools, which is the nations 22nd-largest school district and serves more than 100,000 students. While the district investigated, Mr. Eiswert, who denied making the comments, was inundated with threats to his safety, the police said. He was also placed on administrative leave, the school district said.

Now Mr. Darien is facing charges including disrupting school operations and stalking the principal.

Mr. Eiswert referred a request for comment to a trade group for principals, the Council of Administrative and Supervisory Employees, which did not return a call from a reporter. Mr. Darien, who posted bond on Thursday, could not immediately be reached for comment.

The Baltimore County case is just the latest indication of an escalation of A.I. abuse in schools. Many cases include deepfakes, or digitally altered video, audio or images that can appear convincingly real.

Since last fall, schools across the United States have been scrambling to address troubling deepfake incidents in which male students used A.I. nudification apps to create fake unclothed images of their female classmates, some of them middle school students as young as 12. Now the Baltimore County deepfake voice incident points to another A.I. risk to schools nationwide this time to veteran educators and district leaders.

We are having trouble retrieving the article content.

Please enable JavaScript in your browser settings.

Thank you for your patience while we verify access. If you are in Reader mode please exit andlog intoyour Times account, orsubscribefor all of The Times.

Thank you for your patience while we verify access.

Already a subscriber?Log in.

Want all of The Times?Subscribe.

Read the rest here:

Racist AI Deepfake of Baltimore Principal Leads to Arrest - The New York Times

AI deepfakes are already hitting elections. We have little protection. – The Washington Post

Divyendra Singh Jadouns phone is ringing off the hook. Known as the Indian Deepfaker, Jadoun is famous for using artificial intelligence to create Bollywood sequences and TV commercials.

But as staggered voting in Indias election begins, Jadoun says hundreds of politicians have been clamoring for his services, with more than half asking for unethical things. Candidates asked him to fake audio of competitors making gaffes on the campaign trail or to superimpose challengers faces onto pornographic images. Some campaigns have requested low-quality fake videos of their own candidate, which could be released to cast doubt on any damning real videos that emerge during the election.

Jadoun, 31, says he declines jobs meant to defame or deceive. But he expects plenty of consultants will oblige, bending reality in the worlds largest election, as more than half a billion Indian voters head to the polls.

The only thing stopping us from creating unethical deepfakes is our ethics, Jadoun told The Washington Post. But its very difficult to stop this.

Indias elections, which began last week and run until early June, offer a preview of how an explosion of AI tools is transforming the democratic process, making it easy to develop seamless fake media around campaigns. More than half the worlds population lives in the more than 50 countries hosting elections in 2024, marking a pivotal year for global democracies.

While its unknown how many AI fakes have been made of politicians, experts say they are observing a global uptick of electoral deepfakes.

I am seeing more [political deepfakes] this year than last year and the ones I am seeing are more sophisticated and compelling, said Hany Farid, a computer science professor at the University of California at Berkeley.

While policymakers and regulators from Brussels to Washington are racing to craft legislation restricting the use of AI-powered audio, images and videos on the campaign trail, a regulatory vacuum is emerging. The European Unions landmark AI Act doesnt take effect until after June parliamentary elections. In the U.S. Congress, bipartisan legislation that would ban falsely depicting federal candidates using AI is unlikely to become law before the November elections. A handful of U.S. states have enacted laws penalizing people who make deceptive videos about politicians, creating a patchwork of policy across the country.

In the meantime, there are limited guardrails to deter politicians and their allies from using AI to dupe voters, and enforcers are rarely a match for fakes that can spread quickly across social media or in group chats. The democratization of AI means its up to individuals like Jadoun not regulators to make ethical choices to stave off AI-induced election chaos.

Lets not stand on the sidelines while our elections get screwed up, Sen. Amy Klobuchar (D-Minn.), who chairs the Senate Rules Committee, said in a speech last month at the Atlantic Council. This is like a hair on fire moment. This is not a lets wait three years and see how it goes moment.

For years, nation-state groups flooded Facebook, Twitter (now X) and other social media with misinformation, emulating the playbook Russia famously used in 2016 to stoke discord in U.S. elections. But AI allows smaller actors to partake, making combating falsehoods a fractured and difficult undertaking.

The Department of Homeland Security warned election officials in a memo that generative AI could be used to enhance foreign-influence campaigns targeting elections. AI tools could allow bad actors to impersonate election officials, DHS said in the memo, spreading incorrect information about how to vote or the integrity of the election process.

These warnings are becoming a reality across the world. State-backed actors used generative AI to meddle in Taiwans elections earlier this year. On election day, a Chinese Communist Party-affiliated group posted AI-generated audio of a prominent politician who dropped out of the Taiwanese election throwing his support behind another candidate, according to a Microsoft report. But the politician, Foxconn owner Terry Gou, had never made such an endorsement, and YouTube pulled down the audio.

Divyendra Singh Jadoun used AI to morph Indian Prime Minister Modis voice into making personalized greetings for the Hindu holiday of Diwali. (Video: Divyendra Singh Jadoun)

Taiwan ultimately elected Lai Ching-te, a candidate whom Chinese Communist Party leadership opposed signaling the limits of the campaign to affect the results of the election.

Microsoft expects China to use a similar playbook in India, South Korea and the United States this year. Chinas increasing experimentation in augmenting memes, videos, and audio will likely continue and may prove more effective down the line, the Microsoft report said.

But the low cost and broad availability of generative AI tools have made it possible for people without state backing to engage in trickery that rivals nation-state campaigns.

In Moldova, AI deepfake videos have depicted the countrys pro-Western president, Maia Sandu, resigning and urging people to support a pro-Putin party during local elections. In South Africa, a digitally altered version of the rapper Eminem endorsed a South African opposition party ahead of the countrys election in May.

In January, a Democratic political operative faked President Bidens voice to urge New Hampshire primary voters to not go to the polls a stunt intended to draw awareness to the problems with the medium.

The rise of AI deepfakes could shift the demographics of who runs for office, because bad actors disproportionately use synthetic content to target women.

For years, Rumeen Farhana, an opposition-party politician in Bangladesh, has faced sexual harassment on the internet. But last year, an AI deepfake photo of her in a bikini emerged on social media.

Farhana said it is unclear who made the image. But in Bangladesh, a conservative majority-Muslim country, the photo drew harassing comments from ordinary citizens on social media, with many voters assuming the photo was real.

Such character assassinations might prevent female candidates from subjecting themselves to political life, Farhana said.

Whatever new things come up, its always used against the women first. They are the victim in every case, Farhana said. AI is not an exception in any way.

In the absence of activity from Congress, states are taking action while international regulators are inking voluntary commitments from companies.

About 10 states have adopted laws that would penalize those who use AI to dupe voters. Last month, Wisconsins governor signed a bipartisan bill into law that would fine people who fail to disclose AI in political ads. And a Michigan law punishes anyone who knowingly circulates an AI-generated deepfake within 90 days of an election.

Yet its unclear if the penalties which include fines of up to $1,000 and up to 90 days of jail time, depending on the municipality are steep enough to deter potential offenders.

With limited detection technology and few designated personnel, enforcers may find it difficult to quickly confirm if a video or image is actually AI-generated.

In the absence of regulations, government officials are seeking voluntary agreements from politicians and tech companies alike to control the proliferation of AI-generated election content. European Commission Vice President Vera Jourova said she has sent letters to key political parties in European member states with a plea to resist using manipulative techniques. However, she said, politicians and political parties will face no consequences if they do not heed her request.

I cannot say whether they will follow our advice or not, she said in an interview. I will be very sad if not because if we have the ambition to govern in our member states, then we should also show we can win elections without dirty methods.

Jourova said that in July 2023 she asked large social media platforms to label AI-generated productions ahead of the elections. The request received a mixed response in Silicon Valley, where some platforms told her it would be impossible to develop technology to detect AI.

OpenAI, which makes the chatbot ChatGPT and image generator DALL-E, has also sought to form relationships with the social media companies to address the distribution of AI-generated political materials. At the Munich Security Conference in February, 20 leading technology companies pledged to team up to detect and remove harmful AI content during the 2024 elections.

This is a whole-of-society issue, Anna Makanju, OpenAI vice president of global affairs, said during a Post Live interview. It is not in any of our interests for this technology to be leveraged in this way, and everyone is quite motivated, particularly because we now have lessons from prior elections and from prior years.

Yet companies will not face any penalties if they fail to live up to their pledge. Already, there have been gaps between OpenAIs stated policies and its enforcement. A super PAC backed by Silicon Valley insiders launched an AI chatbot of long-shot presidential candidate Dean Phillips powered by the companys ChatGPT software, in violation of OpenAIs prohibition political campaigns use of its technology. The company did not ban the bot until The Post reported on it.

Jadoun, who does AI political work for Indias major electoral parties, said the spread of deepfakes cant be solved by government alone citizens must be more educated.

Any content that is making your emotions rise to a next level, he said, just stop and wait before sharing it.

Excerpt from:

AI deepfakes are already hitting elections. We have little protection. - The Washington Post

The Week’s 10 Biggest Funding Rounds: Xaira And Other AI Startups Have Huge Week – Crunchbase News

Want to keep track of the largest startup funding deals in 2024 with our curated list of $100 million-plus venture deals to U.S.-based companies? Check out The Crunchbase Megadeals Board.

This is a weekly feature that runs down the weeks top 10 announced funding rounds in the U.S. Check out last weeks biggest funding rounds here.

After a slowdown in big rounds last week, investors were back at it again dishing out nine-figure rounds. This weeks theme was definitely AI, as many of the largest rounds went to startups using AI in coding or biotech.

1. Xaira Therapeutics, $1B, biotech: The biggest round this week was really big. Xaira Therapeutics came out of stealth and announced it had secured more than $1 billion of committed capital from lead investors Arch Venture Partners and Foresite Capital both of which jointly incubated the company as well as several other big-name investors including Sequoia Capital and Lightspeed Venture Partners. The San Francisco-based biotech firm is led by founding CEO Marc Tessier-Lavigne, who previously served as president of Stanford University but resigned last year after questions arose concerning his scientific research. Xaira is the latest although likely best-funded startup to try to use AI models to find new drugs.

2. Augment, $227M, artificial intelligence: AI coding startups made a big splash this week. The biggest round went to another startup emerging from stealth AI coding assistance startup Augment. The Palo Alto, California-based company locked up a $227 million Series B round at a $977 million post-money valuation. The round included cash from the likes of Sutter Hill Ventures, Index Ventures and Lightspeed Venture Partners. Augment helps developers and software teams by giving them AI coding assistance. Founded in 2022, Augments total funding now stands at $252 million, per the company, following its $25 million Series A led by Sutter Hill Ventures in January.

3. Rippling, $200M, human resources: HR startup Rippling wrapped up a huge $200 million round that valued the San Francisco-based startup at $13.5 billion. The deal also included a $590 million secondary offering for employees to sell their private shares to investors. The new equity round was led by Coatue. The company offers a workforce management platform that combines HR, information technology and finance to help customers streamline operations. The new round represents about a 20% uptick in valuation for Rippling, which was last valued at $11.3 billion in March 2023.

4. Cognition, $175M, artificial intelligence: Another nine-figure round for an AI coding startup. This one went to San Francisco-based Cognition, which reportedly locked up a $175 million investment led by Founders Fund at a $2 billion valuation. The six-month-old startup has developed an artificial intelligencepowered coding assistant called Devin. Just last month the startup raised a $21 million Series A at a $350 million valuation.

5. Endeavor BioMedicines, $133M, biotech: Yet another big biotech raise this week. San Diego-based Endeavor BioMedicines locked up a $132.5 million Series C led by AyurMaya. The company, which focuses on medicines for idiopathic pulmonary fibrosis, last raised more than two years ago a $101 million Series B co-led by Ally Bridge Group and Avidity Partners. Founded in 2018, the company has raised nearly $296 million, per Crunchbase.

6. ThreatLocker, $115M, cybersecurity: Cyber funding has shown some signs of life recently, and this week we have another example. ThreatLocker, a cybersecurity startup offering zero trust endpoint security solutions, raised a $115 million Series D led by existing investor General Atlantic. In the past year, ThreatLocker has doubled its revenue and added 50% to its workforce. Founded in 2017, the company has raised nearly $240 million, per Crunchbase.

7. Lumeris, $100M, healthcare: St. Louis-based Lumeris, a developer of a healthcare spending platform, raised a $100 million investment led by Deerfield Management and Endeavor Health. Founded in 2000, the company has raised $325 million, per Crunchbase.

8. Perplexity AI, $63M, artificial intelligence: It was a busy week for Perplexity AI. The company announced a $62.7 million round on the same day it was reported the AI startup was looking to raise another $250 million-plus at a valuation between $2.5 billion and $3 billion. The newly announced round was first reported last month and was led by Daniel Gross. It also included investors such as Nvidia, IVP, NEA, Jeff Bezos and Garry Tan among others. The new round reportedly values the company at more than $1 billion. However, per the report in TechCrunch, the AI search engine startup is far from done, engaging investors in talks of a megaround that would increase that valuation by at least 150%. IVP and NEA were said to be looking at participating in the new round. It was just in January the company raised a $73.6 million Series B led by IVP that valued it at $520 million.

9. Midi Health, $60M, healthcare: Los Altos, California-based Midi Health, a virtual care clinic for women, closed a $60 million Series B led by Emerson Collective. Founded in 2021, Midi Health has raised $100 million to date, per the company.

10. Givebutter, $50M, CRM: Austin, Texas-based Givebutter, a platform for nonprofit fundraising and CRM, announced it has closed a $50 million strategic growth investment led by BVP Forge. Founded in 2016, the company has raised $57 million, per Crunchbase.

There were not really any large rounds outside the U.S. this week. The biggest came from the Emerald Isle.

We tracked the largest announced rounds in the Crunchbase database that were raised by U.S.-based companies for the seven-day period of April 20 to 26. Although most announced rounds are represented in the database, there could be a small time lag as some rounds are reported late in the week.

Illustration: Dom Guzman

Stay up to date with recent funding rounds, acquisitions, and more with the Crunchbase Daily.

View original post here:

The Week's 10 Biggest Funding Rounds: Xaira And Other AI Startups Have Huge Week - Crunchbase News

A National Security Insider Does the Math on the Dangers of AI – WIRED

One type of risk youve been very interested in for a long time is biorisk. Whats the worst thing that could possibly happen? Take us through that.

I started out in public health before I worked in national security, working on infectious disease controlmalaria and tuberculosis. In 2002, the first virus was synthesized from scratch on a Darpa project, and it was sort of an oh crap moment for the biosciences and the public health community, realizing biology is going to become an engineering discipline that could be potentially misused. I was working with veterans of the smallpox eradication campaign, and they thought, Crap, we just spent decades eradicating a disease that now could be synthesized from scratch.

There's a lot of vulnerability in society. Covid was a demonstration of this.

Jason Matheny

I then moved to working in biosecurity, trying to figure out, How could we increase the security around biolabs so that they're less likely to be used? How can we detect biological weapons programs? Which, unfortunately, still exist in significant numbers in a few places in the world. Also, how can we bake more security into society so that we're more resilient when it comes not only to an engineered pandemic but also natural pandemics?

There's a lot of vulnerability that remains in society. Covid was a demonstration of this. This was a relatively mild virus in historic termsit had an infection fatality rate less than 1 percentwhereas there are some natural viruses that have fatality rates well above 50 percent. There are synthetic viruses that have close to 100 percent lethality while still being as transmissible as SARS-CoV-2. Even though we know how to design vaccines and manufacture them very quickly, getting them approved takes about as much time today as it did about 20 years ago. So the amount of time that you would need in order to vaccinate a population is about the same today as it was for our parents and even for our grandparents.

When I first started getting interested in biosecurity in 2002, it cost many millions of dollars to construct a poliovirus, a very, very small virus. It would've cost close to $1 billion to synthesize a pox virus, a very large virus. Today, the cost is less than $100,000, so it's a 10,000-fold decrease over that period. Meanwhile, vaccines have actually tripled in cost over that period. The defense-offense asymmetry is moving in the wrong direction.

And what do you see as our greatest adversary in biorisks?

First is nature. The evolution of natural viruses continues. We're going to have future viral pandemics. Some of them are going to be worse than Covid, some of them are going to be not as bad as Covid, but we've got to be resilient to both. Covid cost just the US economy more than $10 trillion, and yet what we invest in preventing the next pandemic is maybe $2 billion to $3 billion of federal investment.

Another category is intentional biological attacks. Aum Shinrikyo was a doomsday cult in Japan that had a biological weapons program. They believed that they would be fulfilling prophecy by killing everybody on the planet. Fortunately, they were working with 1990s biology, which wasn't that sophisticated. Unfortunately, they then turned to chemical weapons and launched the Tokyo sarin gas attacks.

The barrier to entry for somebody who wants to carry out a biological attack is eroding.

Jason Matheny

We have individuals and groups today that have mass-casualty intent and increasingly express interest in biology as a weapon. What's preventing them from being able to use biology effectively are not controls on the tools or the raw materials, because those are all now available in many laboratories and on eBayyou can buy a DNA synthesizer for much less than $100,000 now. You can get all the materials and consumables that you need from most scientific supply stores.

What an apocalyptic group would lack is the know-how to turn those tools into a biological weapon. Theres a concern that AI makes the know-how more widely available. Some of the research done by [AI safety and research company] Anthropic has looked at risk assessments to see if these tools could be misused by somebody who didn't have a strong bio background. Could they basically get graduate-level training from a digital tutor in the form of a large language model? Right now, probably not. But if you map the progress over the last couple of years, the barrier to entry for somebody who wants to carry out a biological attack is eroding.

Original post:

A National Security Insider Does the Math on the Dangers of AI - WIRED

Police: Purported recording of principal contained AI content – WBAL TV Baltimore

ABOUT THIS DISTURBING CASE. IT FEELS LIKE REALLY UNCOMFORTABLE. ITS LIKE THE ENVIRONMENT AT THE SCHOOL JUST STARTS TO FEEL HOSTILE, MORE AND MORE LIKE EVERY DAY IM HERE. PIKESVILLE HIGH SCHOOL STUDENTS ARE FRUSTRATED WITH THE SCANDAL AMONGST THEIR SCHOOL LEADERS. THEYVE ALL HEARD THE AUDIO CLIP THAT SOUNDS LIKE THEYRE NOW REMOVED. PIKESVILLE HIGH SCHOOL PRINCIPAL ERIC EISWERTH. THE CLIP, POSTED TO SOCIAL MEDIA, INCLUDES RACIST AND ANTI-SEMITIC INSULTS ABOUT STUDENTS. NOW, COUNTY LEADERS BELIEVE IT WAS A AI GENERATED FROM THE VERY BEGINNING. WE HAVE EXPRESSED OUR CONCERN ABOUT THE DISTURBING AUDIO THAT SHOOK OUR COMMUNITIES HERE IN BALTIMORE COUNTY, DETECTIVES FOUND FORMER ATHLETIC DIRECTOR 31 YEAR OLD DAYS ON DARIAN TO BE RESPONSIBLE FOR THE CLIP. BELIEVED TO BE RETALIATION. AFTER FINDING OUT HIS CONTRACT WOULD NOT BE RENEWED AFTER TERMINATING A COACH WITHOUT APPROVAL AND ALLEGEDLY PAYING TEACHERS UNDER THE TABLE FOR HELPING HIM WITH HIS JOB, WHICH IS THE SUBJECT OF AN INTERNAL AUDIT. I AM UNABLE TO GET INTO SPECIFIC PERSONNEL DETAILS. I CAN TELL YOU THAT WE ARE EXPLORING ADMINISTRATIVE CHANGES, GIVEN THE IMMENSE IMPACT OF THIS INCIDENT ON THE PIKESVILLE HIGH SCHOOL, SCHOOL COMMUNITY AND THE ONGOING LOCAL, NATIONAL AND INTERNATIONAL COVERAGE. CHARGING DOCUMENT STATE THAT AN EDUCATOR ADMITTED TO POLICE. SHE SENT THE CLIP TO NEWS OUTLETS AND A STUDENT WHO SHE KNEW WOULD SPREAD THE CLIP AROUND. DARIAN ORIGINALLY DENIED INVOLVEMENT, BUT POLICE HAVE NOW CONNECTED THE EMAIL ADDRESS THAT SENT OUT THE AUDIO CLIP TO DARIENS GRANDMOTHERS HOME IN LA, WITH THE RECOVERY PHONE NUMBER BEING DARIENS PHONE. POLICE ALSO HAD MULTIPLE EXPERTS CONFIRMED TRACES OF AI WITH HUMAN EDITING IN THE CLIP, AND FOUND THAT DARIAN HAD EVEN USED THE SCHOOL NETWORK TO SEARCH FOR AI TOOLS. ALL THIS LEADING TO HIS ARREST AT BWI. WHEN ATTEMPTING TO BOARD A FLIGHT THURSDAY MORNING. HES NOW CHARGED WITH THEFT, DISRUPTION OF SCHOOL ACTIVITIES, RETALIATION, AND STALKING. MEANTIME, THE INVESTIGATION INTO THE OTHER EDUCATORS POSSIBLY INVOLVED THAT IS ONGOING AT PIKESVIL

Police: Purported recording of Pikesville HS principal was not authentic, contained AI content

School's former athletic director arrested at BWI-Marshall Airport on open warrant

Updated: 6:08 PM EDT Apr 25, 2024

The former athletic director at Pikesville High School has been arrested in connection to an AI-generated impersonation of the school's principal.Dazhon Darien, 31, faces multiple charges, including stalking, theft, disruption of school operations, and retaliation against a witness, according to charging documents obtained by 11 News.The investigationPolice said the charges are connected to an investigation that started on Jan. 17 into a voice recording circulated on social media that was alleged to have been the voice of the school's principal, Eric Eiswert.Police said the audio clip, which was an alleged race-based commentary on the school's students and teachers, was spread on social media. That led to the temporary removal of Eiswert from the school, as well as many hate-filled messages on social media and numerous calls to the school."From the very beginning, we have expressed our concern about the disturbing audio that shook our communities here in Baltimore County," Baltimore County Executive Johnny Olszewski said.Police: Recording was not authenticPolice said Thursday afternoon that detectives have conclusive evidence based on a forensic analysis by the FBI that the recording was not authentic. Police said the analysis indicated the recording contained traces of AI-generated content. Detectives then sought an additional analysis by the University of California, Berkley, which arrived at the same findings as the FBI's analysis.Police said investigators believe Darien made the recording to retaliate against Eiswert, who, at the time, was pursuing an investigation into the potential mishandling of school funds. From the day the audio clip went out, Eiswert told detectives he had an idea of who was behind it. The charging documents state that Eiswert believed Darien was responsible for the clip because he believed Darien was tech-savvy and familiar with AI, and a grievance was fueled by Darien's contract not being renewed due to work-performance issues that included terminating a coach without approval and allegedly paying teachers under the table for helping him with his job, which is the subject of an internal audit."While I am unable to get into specific personnel details, we are exploring administrative changes given the immense impact of this incident on the Pikesville High School community and the local, national and international coverage," Baltimore County Public Schools Superintendent Myriam Rogers said.Video below: Full news conferencePolice said Darien had used the school network to search for AI tools. While Darien originally denied involvement, police have since connected an email address that sent out the audio clip to Darien's grandmother's home in Los Angeles that has a recovery phone number being Darien's own cellphone.Others were involvedPolice said the audio clip was originally sent to a friend group of three teachers. One of the educators admitted to police she sent the clip to news outlets and a student who she knew would spread the clip.An investigation into the other teachers involved continues.Baltimore County leaders addressed the scandal in the school district in a news conference Thursday afternoon."As you can imagine, this has been a very difficult time for Pikesville High School community, Principal Eiswert and his family and team BCPS," Rogers said.Pikesville students expressed frustration with the scandal amongst their school leaders."It feels really uncomfortable. The environment at the school starts to feel hostile more and more every day as I'm here," said Julian Solomon, a Pikesville senior.Defendant arrested at airportDarien was apprehended Thursday at Baltimore-Washington International Thurgood Marshall Airport, while he was en route to Houston."Officers with the Maryland Transportation Authority took Mr. Darien into custody this morning at BWI-Thurgood Marshall Airport as he was attempting to board a flight," Baltimore County Police Chief Robert McCullough said.The police chief could not say whether Darien was attempting to flee."He could have been simply going there for other purposes," McCullough said.The chief did say Darien had a firearm, which alerted Maryland Transportation Authority police because of the way the firearm was packaged for transport. MDTA police then detained Darien after discovering the warrant out for his arrest in the Pikesville case. Prosecutor: This appears to be a first-of-its-kind caseBaltimore County State's Attorney Scott Shellenberger said the case appears to one of the first of its kind nationwide that his office was able to find. He said the Maryland General Assembly may need to update state laws to catch up with the nefarious possibilities of the new technology. For example, the charge of disrupting school activities "only carries a six-month sentence," Shellenberger said. "But we also need to take a broader look at how this technology can be used and abused to harm other people."Following a court appearance Thursday afternoon, Darien was released on a $5,000 unsecured bond. A check of online court records for Darien did not list an attorney who might be able to speak on his behalf Thursday.BCPS statementBaltimore County Public Schools sent an email to its community, which follows below.I am writing to provide an important update on the Pikesville High School investigation. This morning, Baltimore County police officers arrested Dazhon Darien, the Pikesville High School athletic director, on multiple charges stemming from an investigation the Baltimore County Police Department launched in reference to the Jan. 17, 2024, release of an audio recording allegedly of Principal Eric Eiswert. This afternoon, I joined the county executive, chief of police, and the state's attorney for a joint press conference so that the police could provide an update on the investigation and I could address next steps for the Pikesville High School community and Team BCPS. When we were made aware of the recording, we immediately launched an investigation and solicited the help of the county executive, Baltimore County police, and additional external experts. Baltimore County police enlisted the services of the Federal Bureau of Investigation. We collaborated with these partners because they have the resources and skill to determine the authenticity of the recording, and they have verified that Mr. Darien used artificial intelligence to create the audio recording. In light of today's announcement by the police department and as required by state education law we are taking appropriate action regarding Mr. Darien's conduct, up to and including a recommendation for termination. We have also been made aware that other BCPS employees are named in the charging documents. The BCPS Office of Human Resources will follow all system procedures for investigating their involvement and we will take swift action in alignment with their rights under the established collective bargaining agreement, board policy and state law.We want to thank County Executive Johnny Olszewski, Chief of Police Robert McCullough, State's Attorney Scott Shellenberger and our law enforcement partners for their work over the last several months. While law enforcement investigated, our focus as a system remained on providing direct support to Pikesville High School students and staff and creating safe spaces for them to process this incident, while maintaining our unwavering commitment to teaching and learning. This has been a difficult time for the Pikesville High School community, Principal Eiswert and his family and Team BCPS. We are proud of the students and staff and how they have stepped up to support one another. As I stated previously, the statements that were made in the recording do not reflect the core values of our system. We are proud of the diversity of Baltimore County Public Schools and strive to ensure that each member of Team BCPS feels valued and has a sense of belonging.We can now shift our focus to moving forward and providing all parties with a fresh start. While we are unable to provide specific personnel-related details, it is important to note that we are exploring administrative changes given the immense impact of this incident on the Pikesville High School community and the ongoing local, national and international coverage. To ensure continuity for our students, Mrs. Kyria Joseph, executive director of high schools, and Dr. George Roberts, consulting administrator, will continue to serve the Pikesville High School community as lead administrators for the remainder of the school year. We will work with Principal Eiswert and the Pikesville community concerning next school year.BCPS student support services team members will be available Friday, April 25, 2024, and on Monday, April 28, 2024, to engage Pikesville High School students in conversations as they process this information. We also encourage parents and families to engage in an open dialogue with their children.Resources to support parents and families include:Assisting parents caregivers in coping with collective traumas (PDF)Resources that are available to BCPS staff and the community:BCPS Employee Assistance ProgramTalkspace: Free online therapy for high school students in BaltimoreBaltimore County Crisis Response: Call 410-931-2214National Suicide Prevention Lifeline: Call 988Crisis text line: Text the word "HOME" to 741741 for free 24-hour support211 Maryland: Call 211We also want to encourage Pikesville High School students to speak with school support staff (i.e., counselors, social worker, psychologist) and trusted adults. Parents and caretakers are welcome to reach out to staff if they need support for their student. Thank you for your patience and support as we worked through the investigative process with law enforcement. We hope that today's update can provide some sense of closure to the community, and we look forward to our continued partnership to ensure the Pikesville High School community and Team BCPS end this school year strong.The Associated Press contributed to this report.

The former athletic director at Pikesville High School has been arrested in connection to an AI-generated impersonation of the school's principal.

Dazhon Darien, 31, faces multiple charges, including stalking, theft, disruption of school operations, and retaliation against a witness, according to charging documents obtained by 11 News.

Police said the charges are connected to an investigation that started on Jan. 17 into a voice recording circulated on social media that was alleged to have been the voice of the school's principal, Eric Eiswert.

Police said the audio clip, which was an alleged race-based commentary on the school's students and teachers, was spread on social media. That led to the temporary removal of Eiswert from the school, as well as many hate-filled messages on social media and numerous calls to the school.

"From the very beginning, we have expressed our concern about the disturbing audio that shook our communities here in Baltimore County," Baltimore County Executive Johnny Olszewski said.

Police said Thursday afternoon that detectives have conclusive evidence based on a forensic analysis by the FBI that the recording was not authentic. Police said the analysis indicated the recording contained traces of AI-generated content. Detectives then sought an additional analysis by the University of California, Berkley, which arrived at the same findings as the FBI's analysis.

Police said investigators believe Darien made the recording to retaliate against Eiswert, who, at the time, was pursuing an investigation into the potential mishandling of school funds.

From the day the audio clip went out, Eiswert told detectives he had an idea of who was behind it. The charging documents state that Eiswert believed Darien was responsible for the clip because he believed Darien was tech-savvy and familiar with AI, and a grievance was fueled by Darien's contract not being renewed due to work-performance issues that included terminating a coach without approval and allegedly paying teachers under the table for helping him with his job, which is the subject of an internal audit.

"While I am unable to get into specific personnel details, we are exploring administrative changes given the immense impact of this incident on the Pikesville High School community and the local, national and international coverage," Baltimore County Public Schools Superintendent Myriam Rogers said.

Video below: Full news conference

Police said Darien had used the school network to search for AI tools. While Darien originally denied involvement, police have since connected an email address that sent out the audio clip to Darien's grandmother's home in Los Angeles that has a recovery phone number being Darien's own cellphone.

Police said the audio clip was originally sent to a friend group of three teachers. One of the educators admitted to police she sent the clip to news outlets and a student who she knew would spread the clip.

An investigation into the other teachers involved continues.

Baltimore County leaders addressed the scandal in the school district in a news conference Thursday afternoon.

"As you can imagine, this has been a very difficult time for Pikesville High School community, Principal Eiswert and his family and team BCPS," Rogers said.

Pikesville students expressed frustration with the scandal amongst their school leaders.

"It feels really uncomfortable. The environment at the school starts to feel hostile more and more every day as I'm here," said Julian Solomon, a Pikesville senior.

Darien was apprehended Thursday at Baltimore-Washington International Thurgood Marshall Airport, while he was en route to Houston.

"Officers with the Maryland Transportation Authority took Mr. Darien into custody this morning at BWI-Thurgood Marshall Airport as he was attempting to board a flight," Baltimore County Police Chief Robert McCullough said.

The police chief could not say whether Darien was attempting to flee.

"He could have been simply going there for other purposes," McCullough said.

The chief did say Darien had a firearm, which alerted Maryland Transportation Authority police because of the way the firearm was packaged for transport. MDTA police then detained Darien after discovering the warrant out for his arrest in the Pikesville case.

Baltimore County State's Attorney Scott Shellenberger said the case appears to one of the first of its kind nationwide that his office was able to find. He said the Maryland General Assembly may need to update state laws to catch up with the nefarious possibilities of the new technology.

For example, the charge of disrupting school activities "only carries a six-month sentence," Shellenberger said. "But we also need to take a broader look at how this technology can be used and abused to harm other people."

Following a court appearance Thursday afternoon, Darien was released on a $5,000 unsecured bond. A check of online court records for Darien did not list an attorney who might be able to speak on his behalf Thursday.

Baltimore County Public Schools sent an email to its community, which follows below.

Resources to support parents and families include:

Resources that are available to BCPS staff and the community:

The Associated Press contributed to this report.

Excerpt from:

Police: Purported recording of principal contained AI content - WBAL TV Baltimore

Adobe’s impressive AI upscaling project makes blurry videos look HD – The Verge

Adobe researchers have developed a new generative AI model called VideoGigaGAN that can upscale blurry videos at up to eight times their original resolution. Introduced in a paper published on April 18th, Adobe claims VideoGigaGAN is superior to other Video Super Resolution (VSR) methods as it can provide more fine-grained details without introducing any AI weirdness to the footage.

In a nutshell, Generative Adversarial Networks (GANs) are effective for upscaling still images to a higher resolution, but struggle to do the same for video without introducing flickering and other unwanted artifacts. Other upscaling methods can avoid this, but the results arent as sharp or detailed. VideoGigaGAN aims to provide the best of both worlds the higher image/video quality of GAN models, with fewer flickering or distortion issues across output frames. The company has provided several examples here that show its work in full resolution.

Some of the finer details in the demo clips Adobe provided appear to be entirely artificial, such as the skin texture and creases in the below example, but the results appear impressively natural. It would be difficult to tell that generative AI was used to improve the resolution, which could extend the what is a photo debate to include video.

This is only a research preview, so theres no guarantee that Adobe will make VideoGigaGAN available to consumers via Creative Cloud software like Premiere Pro. The company previously previewed a separate diffusion-based upsampling experiment, Project Res-Up, during its MAX event in October 2023, which similarly improves the quality of low-resolution GIFs and video footage. And Adobe isnt alone in this work, as both Microsoft and Nvidia have also developed their own VSR upscaling technology.

Read this article:

Adobe's impressive AI upscaling project makes blurry videos look HD - The Verge