In a huge move for the mixed reality industry, Meta announced today that it's opening the Quest's operating system to third-party companies, allowing them to build headsets of their own. From a report: Think of it like moving the Quest's ecosystem from an Apple model, where one company builds both the hardware and software, to more of a hardware free-for-all like Android. The Quest OS is being rebranded to "Meta Horizon OS," and at this point it seems to have found two early adopters. ASUS's Republic of Gamers (ROG) brand is working on a new "performance gaming" headsets, while Lenovo is working on devices for "productivity, learning and entertainment." (Don't forget, Lenovo also built the poorly-received Oculus Rift S.)

As part of the news, Meta says it's also working on a limited-edition Xbox "inspired" Quest headset. (Microsoft and Meta also worked together recently to bring Xbox cloud gaming to the Quest.) Meta is also calling on Google to bring over the Google Play 2D app store to Meta Horizon OS. And, in an effort to bring more content to the Horizon ecosystem, software developed through the Quest App Lab will be featured in the Horizon Store. The company is also developing a new spatial framework to let mobile developers created mixed reality apps.

A military court in Moscow on Monday sentenced Meta spokesperson Andy Stone to six years in prison for "publicly defending terrorism," a verdict handed down in absentia, RIA news agency reported. Reuters: Meta itself is designated an extremist organisation in Russia and its Facebook and Instagram social media platforms have been banned in the country since 2022 when Russia invaded Ukraine.

[...] Russia's interior ministry opened a criminal investigation into Stone late last year, without disclosing specific charges. RIA cited state investigators as saying Stone had published online comments that defended "aggressive, hostile and violent actions" towards Russian soldiers involved in what Moscow calls its "special military operation" in Ukraine.

The EU's European Data Protection Board oversees its privacy-protecting GDPR policies.

Earlier this week, TechCrunch reported that nearly two dozen civil society groups and nonprofits wrote the Board an open letter "urging it not to endorse a strategy used by Meta that they say is intended to bypass the EU's privacy protections for commercial gain."

Meta's strategy is sometimes called "Pay or Okay," writes long-time Slashdot reader AmiMoJo : Meta offers users a choice: "consent" to tracking, or pay over €250/year to use its sites without invasive monetization of personal data.
Meta prefers the phrase "subsccription for no ads," and told TechCrunch it makes them compliant with EU laws: A raft of complaints have been filed against Meta's implementation of the pay-or-consent tactic since it launched the "no ads" subscription offer last fall. Additionally, in a notable step last month, the European Union opened a formal investigation into Meta's tactic, seeking to find whether it breaches obligations that apply to Facebook and Instagram under the competition-focused Digital Markets Act. That probe remains ongoing.
The letter to the Board called for "robust protections that prioritize data subjects' agency and control over their information." And Wednesday the board issued its first decision:

"[I]n most cases, it will not be possible for [social media services] to comply with the requirements for valid consent, if they confront users only with a choice between consenting to processing of personal data for behavioural advertising purposes and paying a fee." The EDPB considers that offering only a paid alternative to services which involve the processing of personal data for behavioural advertising purposes should not be the default way forward for controllers. When developing alternatives, large online platforms should consider providing individuals with an 'equivalent alternative' that does not entail the payment of a fee. If controllers do opt to charge a fee for access to the 'equivalent alternative', they should give significant consideration to offering an additional alternative. This free alternative should be without behavioural advertising, e.g. with a form of advertising involving the processing of less or no personal data.
EDPB Chair, Anu Talus added: "Controllers should take care at all times to avoid transforming the fundamental right to data protection into a feature that individuals have to pay to enjoy."

An anonymous reader quotes a report from Reuters: The Dutch privacy watchdog AP on Friday said it was recommending that government organizations should stop using Facebook as long as it is unclear what happens with personal data of users of the government's Facebook pages. "People that visit a government's page need to be able to trust that their personal and sensitive data is in safe hands," AP chairman Aleid Wolfsen said in a statement. Junior minister for digitalization Alexandra van Huffelen said Facebook parent company Meta had to make clear before the summer how it could take away the government's concerns on the safety of data. "Otherwise we will be forced to stop using Facebook, in line with this advice," she said.

An anonymous reader quotes a report from The Atlantic: In October 2003, Mark Zuckerberg created his first viral site: not Facebook, but FaceMash. Then a college freshman, he hacked into Harvard's online dorm directories, gathered a massive collection of students' headshots, and used them to create a website on which Harvard students could rate classmates by their attractiveness, literally and figuratively head-to-head. The site, a mean-spirited prank recounted in the opening scene of The Social Network, got so much traction so quickly that Harvard shut down his internet access within hours. The math that powered FaceMash -- and, by extension, set Zuckerberg on the path to building the world's dominant social-media empire -- was reportedly, of all things, a formula for ranking chess players: the Elo system.

Fundamentally, what an Elo rating does is predict the outcome of chess matches by assigning every player a number that fluctuates based purely on performance. If you beat a slightly higher-ranked player, your rating goes up a little, but if you beat a much higher-ranked player, your rating goes up a lot (and theirs, conversely, goes down a lot). The higher the rating, the more matches you should win. That is what Elo was designed for, at least. FaceMash and Zuckerberg aside, people have deployed Elo ratings for many sports -- soccer, football, basketball -- and for domains as varied as dating, finance, and primatology. If something can be turned into a competition, it has probably been Elo-ed. Somehow, a simple chess algorithm has become an all-purpose tool for rating everything. In other words, when it comes to the preferred way to rate things, Elo ratings have the highest Elo rating. [...]

Elo ratings don't inherently have anything to do with chess. They're based on a simple mathematical formula that works just as well for any one-on-one, zero-sum competition -- which is to say, pretty much all sports. In 1997, a statistician named Bob Runyan adapted the formula to rank national soccer teams -- a project so successful that FIFA eventually adopted an Elo system for its official rankings. Not long after, the statistician Jeff Sagarin applied Elo to rank NFL teams outside their official league standings. Things really took off when the new ESPN-owned version of Nate Silver's 538 launched in 2014 and began making Elo ratings for many different sports. Some sports proved trickier than others. NBA basketball in particular exposed some of the system's shortcomings, Neil Paine, a stats-focused sportswriter who used to work at 538, told me. It consistently underrated heavyweight teams, for example, in large part because it struggled to account for the meaninglessness of much of the regular season and the fact that either team might not be trying all that hard to win a given game. The system assumed uniform motivation across every team and every game. Pretty much anything, it turns out, can be framed as a one-on-one, zero-sum game. Arpad Emmerich Elo, creator of the Elo rating system, understood the limitations of his invention. "It is a measuring tool, not a device of reward or punishment," he once remarked. "It is a means to compare performances, assess relative strength, not a carrot waved before a rabbit, or a piece of candy given to a child for good behavior."

An anonymous reader shares a report: Today Meta unleashed its ChatGPT competitor, Meta AI, across its apps and as a standalone. The company boasts that it is running on its latest, greatest AI model, Llama 3, which was trained on "data of the highest quality"! A dataset seven times larger than Llama2! And includes 4 times more code! What is that training data? There the company is less loquacious.

Meta said the 15 trillion tokens on which its trained came from "publicly available sources." Which sources? Meta told The Verge that it didn't include Meta user data, but didn't give much more in the way of specifics. It did mention that it includes AI-generated data, or synthetic data: "we used Llama 2 to generate the training data for the text-quality classifiers that are powering Llama 3." There are plenty of known issues with synthetic or AI-created data, foremost of which is that it can exacerbate existing issues with AI, because it's liable to spit out a more concentrated version of any garbage it is ingesting.

WhatsApp users in the U.S. will soon see support for real-time AI image generation. The Verge reports: As soon as you start typing a text-to-image prompt in a chat with Meta AI, you'll see how the image changes as you add more detail about what you want to create. In the example shared by Meta, a user types in the prompt, "Imagine a soccer game on mars." The generated image quickly changes from a typical soccer player to showing an entire soccer field on a Martian landscape. If you have access to the beta, you can try out the feature for yourself by opening a chat with Meta AI and then start a prompt with the word "Imagine."

Additionally, Meta says its Meta Llama 3 model can now produce "sharper and higher quality" images and is better at showing text. You can also ask Meta AI to animate any images you provide, allowing you to turn them into a GIF to share with friends. Along with availability on WhatsApp, real-time image generation is also available to US users through Meta AI for the web. Further reading: Meta Releases Llama 3 AI Models, Claiming Top Performance

Meta debuted a new version of its powerful Llama AI model, its latest effort to keep pace with similar technology from companies like OpenAI, X and Google. The company describes Llama 3 8B and Llama 3 70B, containing 8 billion and 70 billion parameters respectively, as a "major leap" in performance compared to their predecessors.

Meta claims that the Llama 3 models, trained on custom-built 24,000 GPU clusters, are among the best-performing generative AI models available for their respective parameter counts. The company supports this claim by citing the models' scores on popular AI benchmarks such as MMLU, ARC, and DROP, which attempt to measure knowledge, skill acquisition, and reasoning abilities. Despite the ongoing debate about the usefulness and validity of these benchmarks, they remain one of the few standardized methods for evaluating AI models. Llama 3 8B outperforms other open-source models like Mistral's Mistral 7B and Google's Gemma 7B on at least nine benchmarks, showcasing its potential in various domains such as biology, physics, chemistry, mathematics, and commonsense reasoning.

TechCrunch adds: Now, Mistral 7B and Gemma 7B aren't exactly on the bleeding edge (Mistral 7B was released last September), and in a few of benchmarks Meta cites, Llama 3 8B scores only a few percentage points higher than either. But Meta also makes the claim that the larger-parameter-count Llama 3 model, Llama 3 70B, is competitive with flagship generative AI models including Gemini 1.5 Pro, the latest in Google's Gemini series.

Meta plans to "temporarily" shut down Threads in Turkey from April 29, in response to an interim injunction prohibiting data sharing with Instagram. TechCrunch reports: The Turkish Competition Authority (TCA), known as Rekabet Kurumu, noted on March 18 that its investigations found that Meta was abusing its dominant market position by combining the data of users who create Threads profiles with that of their Instagram account -- without giving users the choice to opt in. [...] In the buildup to April 29, everyone using Threads in Turkey will receive a notification about the impending closure, and they will be given a choice to either delete or deactivate their profile. The latter of these options means a user's profile can be resurrected when and if Threads is available in the country again. "We disagree with the interim order, we believe we are in compliance with all Turkish legal requirements, and we will appeal," Meta wrote in the blog post today. "The TCA's interim order leaves us with no choice but to temporarily shut down Threads in Turkiye. We will continue to constructively engage with the TCA and hope to bring Threads back to people in Turkiye as quickly as possible."

An anonymous reader shares a report: Recent headlines have made clear: If AI is doing an impressively good job at a human task, there's a good chance that the task is actually being done by a human. When George Carlin's estate sued the creators of a podcast who said they used AI to create a standup routine in the late comedian's style, the podcasters claimed that the script had actually been generated by a human named Chad. (The two sides recently settled the suit.) A company making AI-powered voice interfaces for fast-food drive-thrus can only complete 30% of jobs without the help of a human reviewing its work. Amazon is dropping its automated "Just Walk Out" checkout systems from new stores -- a system that relied on far more human verification than it was hoping for.

We've seen this before -- though it may already be lost to Silicon Valley's pathologically short memory. Back in 2015, AI chatbots were the hot thing. Tech giants and startups alike pitched them as always-available, always-chipper, always-reliable assistants. One startup, x.ai, advertised an AI assistant who could read your emails and schedule your meetings. Another, GoButler, offered to book your flights or order your fries through a delivery app. Facebook also tested a do-anything concierge service called M, which could answer seemingly any question, do almost any task, and draw you pictures on demand. But for all of those services, the "AI assistant" was often just a person. Back in 2016, I wrote a story about this and interviewed workers whose job it was to be the human hiding behind the bot, making sure the bot never made a mistake or spoke nonsense.

Meta Platforms is planning to launch two small versions of its forthcoming Llama 3 large-language model next week, The Information has reported [non-paywalled link]. From the report: The models will serve as a precursor to the launch of the biggest version of Llama 3, expected this summer. Release of the two small models will likely help spark excitement for the forthcoming Llama 3, which will be coming out roughly a year after Llama 2 launched last July.

It comes as several companies, including Google, Elon Musk's xAI and Mistral, have released open-source LLMs. Meta hopes Llama 3 will catch up with OpenAI's GPT-4, which can answer questions based on images users upload to the chatbot. The biggest version will be multimodal, which means it will be capable of understanding and generating both texts and images. In contrast, the two small models to be released next week won't be multimodal, the employee said.

TechCrunch reports this week that Meta "is denying that it gave Netflix access to users' private messages..." The claim references a court filing that emerged as part of the discovery process in a class-action lawsuit over data privacy practices between a group of consumers and Facebook's parent, Meta. The document alleges that Netflix and Facebook had a "special relationship" and that Facebook even cut spending on original programming for its Facebook Watch video service so as not to compete with Netflix, a large Facebook advertiser. It also says that Netflix had access to Meta's "Inbox API" that offered the streamer "programmatic access to Facebook's user's private message inboxes...."

Meta's communications director, Andy Stone, reposted the original X post on Tuesday with a statement disputing that Netflix had been given access to users' private messages. "Shockingly untrue," Stone wrote on X. "Meta didn't share people's private messages with Netflix. The agreement allowed people to message their friends on Facebook about what they were watching on Netflix, directly from the Netflix app. Such agreements are commonplace in the industry...."

Beyond Stone's X post, Meta has not provided further comment. However, The New York Times had previously reported in 2018 that Netflix and Spotify could read users' private messages, according to documents it had obtained. Meta denied those claims at the time via a blog post titled "Facts About Facebook's Messaging Partnerships," where it explained that Netflix and Spotify had access to APIs that allowed consumers to message friends about what they were listening to on Spotify or watching on Netflix directly from those companies' respective apps. This required the companies to have "write access" to compose messages to friends, "read access" to allow users to read messages back from friends, and "delete access," which meant if you deleted a message from the third-party app, it would also delete the message from Facebook.

"No third party was reading your private messages, or writing messages to your friends without your permission. Many news stories imply we were shipping over private messages to partners, which is not correct," the blog post stated. In any event, Messenger didn't implement default end-to-end encryption until December 2023, a practice that would have made these sorts of claims a non-starter, as it wouldn't have left room for doubt.

Meta is updating its AI-generated content policy and will add a "Made with AI" label beginning next month, the company announced. The policy will apply to content on Instagram, Facebook, and Threads. From a report: Acknowledging that its current policy is "too narrow," Meta says it will start labeling more video, audio, and image content as being AI-generated. Labels will be applied either when users disclose the use of AI tools or when Meta detects "industry standard AI image indicators," though the company didn't provide more detail about its detection system.

The changes are informed by recommendations and feedback from Meta's Oversight Board and update the manipulated media policy created in 2020. The old policy prohibits videos created or edited using AI tools that make a person say something they didn't but doesn't cover the wide range of AI-generated content that has recently flooded the web. "In the last four years, and particularly in the last year, people have developed other kinds of realistic AI-generated content like audio and photos, and this technology is quickly evolving," Meta wrote in a blog post. "As the Board noted, it's equally important to address manipulation that shows a person doing something they didn't do."

Many investors are lining up to bet on the collapse of former President Donald J. Trump's social media company, Trump Media & Technology Group Corp., which made its stock market debut last week under the ticker "DJT." The stock has been called the "mother of all meme stocks" since it is highly volatile and there are no fundamental underpinnings. It's being valued at roughly 1,600 times its annual revenue, at Wednesday's closing price. "By comparison, the stock of Facebook's owner trades at about eight times revenues, and Google's owner trades at six times," notes Fast Company. The New York Times reports: Trump Media is the most "shorted" special purpose acquisition vehicle in the country, according to the financial data company S3 Partners. Short-sellers bet that the price of a stock will fall. They do that by borrowing shares of a company and selling them into the market, hoping to buy them back later at a lower price, before returning the shares to the lender and pocketing the difference as profit. The demand to short Trump Media, the parent company of the social media platform Truth Social, is so great that stock lenders can charge enormous fees, making it hard for short-sellers to turn a profit unless the shares fall significantly. Still, there is a lot of interest in taking the bet. "They are looking for this stock to crater and crater very quickly," said Ihor Dusaniwsky, managing director of predictive analytics at S3. Last month, traders lost $126 million betting against Trump Media, according to S3.

On Monday, Trump Media published updated financial information, revealing little revenue, large losses and a statement from the company's independent auditor expressing "substantial doubt" about its financial viability. This appeared to galvanize investors betting against the company, as the stock slipped from its highs. But short-sellers are finding it difficult and costly to trade in Trump Media. There are roughly 137 million shares in the company, and only around five million of those are available to short-sellers. Mr. Trump owns about 60 percent of shares, and company executives also hold a chunk of the stock. Company insiders tend not to lend their shares to short-sellers. Big asset managers like BlackRock, Vanguard and State Street, which regularly lend out shares, are not major holders of Trump Media, further crimping the supply.

According to S3, 4.9 million of the roughly five million available shares are already on loan. As with any loan, when share owners lend their stock to a short-seller, they charge a fee, usually expressed as an annual interest rate on the stock's current value. Typically, the fee for borrowing stock is a fraction of a percentage point. For Trump Media, it has risen to 550 percent, Mr. Dusaniwsky said. Trump Media's stock currently trades at around $50. That means that shorting it for a month would cost more than $20 per share. For a short-seller to break even, the stock price would have to fall by almost half by early May.

There is another wrinkle, too. One large broker said much of the short trading was not an outright bet against Trump Media. Since the advent of meme-stock trading and the vilification of short-sellers that win only if popular companies lose, large investors are wary of making such trades. Instead, the current trade driving demand is designed to capture the difference between DJT's stock price and outstanding "warrants," which will give the owners the right to new stock at a fixed price as long as regulators approve the new shares. Partly because of that uncertainty, those warrants currently trade below $19, with a list of hedge funds as recent holders. Even after the high cost to borrow stock is accounted for, they are still able to profit from the $30 difference between existing stock and what the warrants are worth, assuming the warrants become registered as shares.

Google Books is indexing low quality, AI-generated books that will turn up in search results, and could possibly impact Google Ngram viewer, an important tool used by researchers to track language use throughout history. From a report: I was able to find the AI-generated books with the same method we've previously used to find AI-generated Amazon product reviews, papers published in academic journals, and online articles. Searching Google Books for the term "As of my last knowledge update," which is associated with ChatGPT-generated answers, returns dozens of books that include that phrase. Some of the books are about ChatGPT, machine learning, AI, and other related subjects and include the phrase because they are discussing ChatGPT and its outputs. These books appear to be written by humans. However, most of the books in the first eight pages of results turned up by the search appear to be AI-generated and are not about AI.

For example, the 2024 book Bears, Bulls, and Wolves: Stock Trading for the Twenty-Year-Old by Tristin McIver, bills itself as "a transformative journey into the world of stock trading" and "a comprehensive guide designed for beginners eager to unlock the mysteries of financial markets." In reality, it reads like ChatGPT-generated text with surface, Wikipedia-level analysis of complex financial events like Facebook's initial public offering or the 2008 financial crisis summed up in a few short paragraphs. [...] Other books appear to be outdated to the point of being useless at the time they are published because they are generated with a version of ChatGPT with an old "knowledge update."

Starting this week, Discord will show ads on the site from video game companies, some of which will offer users gifts for carrying out in-game tasks. According to the Wall Street Journal, Discord said users will be able to turn off the ads in their settings. From a report: The sources said Discord aims to hire more than a dozen ad sales people. WSJ said the addition of ads marks a pivot for Discord, whose CEO Jason Citron has repeatedly said the company would not rely on advertisers the way platforms like Facebook and Instagram do.

The Federal Trade Commission (FTC) has a new way to combat the impersonation scams that it says cost people $1.1 billion last year alone. Effective today, the agency's rule "prohibits the impersonation of government, businesses, and their officials or agents in interstate commerce." The rule also lets the FTC directly file federal court complaints to force scammers to return money stolen by business or government impersonation. From a report: Impersonation scams are wide-ranging -- creators are on the lookout for fake podcast invites that turn into letting scammers take over their Facebook pages via a hidden "datasets" URL, while Verge reporters have been impersonated by criminals trying to steal cryptocurrency via fake Calendly meeting links.

Linus Media Group was victimized by a thief who pretended to be a potential sponsor and managed to take over three of the company's YouTube channels. Some scams can also be very intricate, as in The Cut financial columnist Charlotte Cowles' story of how she lost a shoebox holding $50,000 to an elaborate scam involving a fake Amazon business account, the FTC, and the CIA. (See also: gift card scams.) The agency is also taking public comment until April 30th on changes to the rule that would allow it to also target impersonation of individuals, such as through the use of video deepfakes or AI voice cloning. That would let it take action against, say, scams involving impersonations of Elon Musk on X or celebrities in YouTube ads. Others have used AI for more sinister fraud, such as voice clones of loved ones claiming to be kidnapped.

As a department of America's federal Health agency, the Food and Drug Administration is responsible for public health rules, including prescription medicines. And the FDA "has not changed its position that currently available clinical trial data do not demonstrate that ivermectin is effective against COVID-19," they confirmed to CNN this week. "The agency has not authorized or approved ivermectin for use in preventing or treating COVID-19."

But there was also a lawsuit. In "one of its more popular pandemic-era social media campaigns," the agency tweeted out "You are not a horse. You are not a cow. Seriously, y'all. Stop it." The post attracted nearly 106,000 likes — and over 46,000 reposts, and was followed by another post on Instagram. "Stop it with the #ivermectin. It's not authorized for treating #COVID."

Los Angeles Times business columnist Michael Hiltzik writes that the posts triggered a "groundless" lawsuit: It was those latter two lines that exercised three physicians who had been prescribing ivermectin for patients. They sued the FDA in 2022, asserting that its advisory illegally interfered with the practice of medicine — specifically with their ability to continue prescribing the drug. A federal judge in Texas threw out their case, but the 5th Circuit Court of Appeals — the source of a series of chuckleheaded antigovernment rulings in recent years — reinstated it last year, returning it to the original judge for reconsideration.

Now the FDA has settled the case by agreeing to delete the horse post and two similar posts from its accounts on the social media platforms X, LinkedIn and Facebook. The agency also agreed to retire a consumer advisory titled "Why You Should Not Use Ivermectin to Treat or Prevent COVID-19." In defending its decision, the FDA said it "has chosen to resolve this lawsuit rather than continuing to litigate over statements that are between two and nearly four years old."

That sounds reasonable enough, but it's a major blunder. It leaves on the books the 5th Circuit's adverse ruling, in which a panel of three judges found that the FDA's advisory crossed the line from informing consumers, which they said is all right, to recommending that consumers take some action, which they said is not all right... That's a misinterpretation of the law and the FDA's actions, according to Dorit Rubinstein Reiss of UC College of the Law in San Francisco. "The FDA will seek to make recommendations against the misuse of products in the future, and having that decision on the books will be used to litigate against it," she observed after the settlement.
"A survey by Boston University and the University of Michigan estimated that Medicare and private insurers had wasted $130 million on ivermectin prescriptions for COVID in 2021 alone."

New documents from a class action against Meta "reveal some of the specific ways it tackled rivals in recent years," reports the Observer.

"One of them was using software made by a mobile data analytics company called Onavo in 2016 to access user activities on Snapchat, and eventually Amazon and YouTube, too." Facebook acquired Onavo in 2013 and shut it down in 2019 after a TechCrunch report revealed that the company was paying teenagers to use the software to collect user data.

In 2020, two Facebook users filed a class action lawsuit in the U.S. District Court for the Northern District of California against Meta, then called Facebook, alleging the company engaged in anticompetitive practices and exploited user data. In 2023, the plaintiffs' attorney Brian J. Dunne submitted documents listing how Facebook used Onavo's software to spy on competitors, including Snapchat. According to the documents, made public this week, the Onavo team pitched and launched a project codenamed "Ghostbusters" — in reference to the Snapchat logo — where they developed "kits that can be installed on iOS or Android that intercept traffic for specific sub-domains," allowing them "to read what would otherwise be encrypted traffic so we can measure in-app usage."

The documents also included a presentation from the Onavo team to Mark Zuckerberg showing that they had the ability to track "detailed in-app activity" by "parsing Snapchat analytics collected from incentivized participants in Onavo's program...." The technology was used to do the same to YouTube from 2017 to 2018 and Amazon in 2018, according to the documents. "The intended and actual result of this program was to harm competition, including Facebook's then-nascent Social Advertising competitor Snapchat," the document alleged.

Hillary Clinton and U.S. election officials said they are concerned disinformation generated and spread by AI could threaten the 2024 presidential election [non-paywalled link]. WSJ: Clinton, a former secretary of state and 2016 presidential candidate, said she thinks foreign actors like Russian President Vladimir Putin could use AI to interfere in elections in the U.S. and elsewhere. Dozens of countries are running elections this year. "Anybody who's not worried is not paying attention," Clinton said Thursday at Columbia University, where election officials and tech executives discussed how AI could impact global elections.

She added: "It could only be a very small handful of people in St. Petersburg or Moldova or wherever they are right now who are lighting the fire, but because of the algorithms everyone gets burned." Clinton said Putin tried to undermine her before the 2016 election by spreading disinformation on Facebook, Twitter and Snapchat about "all these terrible things" she purportedly did. "I don't think any of us understood it," she said. "I did not understand it. I can tell you my campaign did not understand it. The so-called dark web was filled with these kinds of memes and stories and videos of all sorts portraying me in all kinds of less than flattering ways." Clinton added: "What they did to me was primitive and what we're talking about now is the leap in technology."