Cloud Hosting

The United States and several allies said Tuesday that they had seized control of a sophisticated Russian propaganda mill that used artificial intelligence to drive nearly a thousand covert accounts on the social network X.

Though governments have increasingly turned to AI in the past year to spread messages more widely and credibly, the takedown is unusual because the Western intelligence agencies traced it to an officer of the Russian FSB intelligence force and to a former senior editor at state-controlled publication RT, formerly called Russia Today, as explained in court filings.

In a strikingly detailed joint advisory, agencies in the United States, the Netherlands and Canada identified various software programs used to manage the network, including one named Meliorator, which created fictitious users known as souls in various countries. The FBI won a court order allowing it to seize two web domains that the operation had used to register the email addresses behind the accounts.

Todays actions represent a first in disrupting a Russian-sponsored Generative AI-enhanced social media bot farm, FBI Director Christopher A. Wray said in a statement. Russia intended to use this bot farm to disseminate AI-generated foreign disinformation, scaling their work with the assistance of AI to undermine our partners in Ukraine and influence geopolitical narratives favorable to the Russian government.

Automated accounts with more detailed biographies posted original content, while a supporting cast of more generic accounts liked and reshared those posts. Officials did not respond to questions about how many real users saw the posts and whether any spread the messages further, so it is unclear how effective the campaign was.

Stories to keep you informed

The system evaded one of Xs techniques for verifying the authenticity of users by automatically copying one-time passcodes sent to the registered email addresses. References to Facebook and Instagram in the program code indicated that the operation intended to expand to those platforms, officials said.

The agencies recommended that social media companies improve their methods for catching covertly automated behavior.

X complied with a court order to furnish information on the accounts to the FBI, then deleted them. The company did not respond to questions from The Washington Post.

The Justice Department thanked X for its cooperation during the investigation, a sign of better communications between the government and the big social media companies after the Supreme Court upheld the right of officials to point out foreign influence operations.

John Scott-Railton, a researcher at the Canadian nonprofit the Citizen Lab, said the countries provided such detailed information about the inner workings of the botnet to help other investigators and companies know what to look for.

They dont think this problem is going anywhere, so they are sharing widely, Scott-Railton said.

The documents show that AIs large language models have helped Russian propagandists scale their operation and help with translation, he said. It also helps them avoid detection software that looks for repeated use of the same internet protocol addresses and other identifiers.

But many other systems are operating already, and they will get better as they adapt for what is getting detected and what is getting by, Scott-Railton said. This isnt even the tip of the iceberg, he said. This is the drip of the iceberg.

Read the rest here:

U.S. and allies take down Russian bot farm powered by AI - The Washington Post

Microsoft has developed a new artificial intelligence (AI) speech generator that is apparently so convincing it cannot be released to the public.

VALL-E 2 is a text-to-speech (TTS) generator that can reproduce the voice of a human speaker using just a few seconds of audio.

Microsoft researchers said VALL-E 2 was capable of generating "accurate, natural speech in the exact voice of the original speaker, comparable to human performance," in a paper that appeared June 17 on the pre-print server arXiv. In other words, the new AI voice generator is convincing enough to be mistaken for a real person at least, according to its creators.

"VALL-E 2 is the latest advancement in neural codec language models that marks a milestone in zero-shot text-to-speech synthesis (TTS), achieving human parity for the first time," the researchers wrote in the paper. "Moreover, VALL-E 2 consistently synthesizes high-quality speech, even for sentences that are traditionally challenging due to their complexity or repetitive phrases."

Related: New AI algorithm flags deepfakes with 98% accuracy better than any other tool out there right now

Human parity in this context means that speech generated by VALL-E 2 matched or exceeded the quality of human speech in benchmarks used by Microsoft.

The AI engine is capable of this given the inclusion of two key features: "Repetition Aware Sampling" and "Grouped Code Modeling."

Get the worlds most fascinating discoveries delivered straight to your inbox.

Repetition Aware Sampling improves the way the AI converts text into speech by addressing repetitions of "tokens" small units of language, like words or parts of words preventing infinite loops of sounds or phrases during the decoding process. In other words, this feature helps vary VALL-E 2's pattern of speech, making it sound more fluid and natural.

Grouped Code Modeling, meanwhile, improves efficiency by reducing the sequence length or the number of individual tokens that the model processes in a single input sequence. This speeds up how quickly VALL-E 2 generates speech and helps manage difficulties that come with processing long strings of sounds.

The researchers used audio samples from speech libraries LibriSpeech and VCTK to assess how well VALL-E 2 matched recordings of human speakers. They also used ELLA-V an evaluation framework designed to measure the accuracy and quality of generated speech to determine how effectively VALL-E 2 handled more complex speech generation tasks.

"Our experiments, conducted on the LibriSpeech and VCTK datasets, have shown that VALL-E 2 surpasses previous zero-shot TTS systems in speech robustness, naturalness, and speaker similarity," the researchers wrote. "It is the first of its kind to reach human parity on these benchmarks."

The researchers pointed out in the paper that the quality of VALL-E 2s output depended on the length and quality of speech prompts as well as environmental factors like background noise.

Despite its capabilities, Microsoft will not release VALL-E 2 to the public due to potential misuse risks. This coincides with increasing concerns around voice cloning and deepfake technology. Other AI companies like OpenAI have placed similar restrictions on their voice tech.

"VALL-E 2 is purely a research project. Currently, we have no plans to incorporate VALL-E 2 into a product or expand access to the public," the researchers wrote in a blog post. "It may carry potential risks in the misuse of the model, such as spoofing voice identification or impersonating a specific speaker."

That said, they did suggest AI speech tech could see practical applications in the future. "VALL-E 2 could synthesize speech that maintains speaker identity and could be used for educational learning, entertainment, journalistic, self-authored content, accessibility features, interactive voice response systems, translation, chatbot, and so on," the researchers added.

They continued: "If the model is generalized to unseen speakers in the real world, it should include a protocol to ensure that the speaker approves the use of their voice and a synthesized speech detection model."

Here is the original post:

AI speech generator 'reaches human parity' but it's too dangerous to release, scientists say - Livescience.com

Palmer Luckey, 31, founder of Anduril Industries, stands in front of the Dive-LD, an autonomous underwater drone at company headquarters in Costa Mesa, Calif. Anduril recently won a U.S. Navy contract to build 200 of them annually. Philip Cheung for NPR hide caption

COSTA MESA, Calif. Its easy to spot Palmer Luckey. Hes the guy with the mullet and the goatee, almost always dressed in khaki shorts, flip flops and bright Hawaiian shirts.

As he gives a tour of Anduril, the artificial intelligence weapons company he founded just south of Los Angeles, hes in his standard business attire.

"This is one of my Dungeons and Dragons Hawaiian shirts," he explains. "You've got an elder dragon. You've got a fighter, a couple of wizards. I wear a lot of Hawaiian shirts because I like them, and I can get away with it.

He can get away with it because he's been a billionaire since his early 20s.

When still a teenager, Luckey launched his first tech company, Oculus, the virtual reality headset for gaming. He sold it a couple years later to Facebook for $2 billion.

Now 31, Luckey took that fortune and founded a new company, Anduril, thats making AI intelligence weapons like drones and submarines.

The Pentagon is buying them, keeping some for itself and sending others to Ukraine. Seven years after it started, Anduril says it's selling its autonomous weapons to about 10 countries worldwide.

In a showroom with its weapons on display, Luckey describes the company's ALTIUS drone.

"It's a drone that fires out of a tube into the air and then unfolds itself, extends its wings, extends its tail, unfolds the propeller and transforms itself into a small airplane," he says. "It can carry up to a 30-pound warhead. So you've got a lot of punch in this thing.

Palmer Luckey walks in the showroom featuring Anduril's artificial intelligence weapons. The company says it can build AI weapons much faster and cheaper than traditional military contractors who make large weapon systems, like fighter jets and tanks. Philip Cheung for NPR hide caption

Anduril is among a growing number of tech companies making artificial intelligence weapons and boldly proclaiming theyll change the way the U.S. and its allies wage war.

In short, the aim is more tech doing the fighting and fewer troops in harm's way. The revolution hasn't happened yet. But these companies are shaking up an industry long dominated by massive firms such as Lockheed Martin and General Dynamics, which build large, traditional weapons, from fighter jets to tanks, designed to last decades.

Anduril, named after a sword in Lord of the Rings, has a very different approach.

"I had this belief that the major defense companies didn't have the right talent or the right incentive structure to invest in things like artificial intelligence, autonomy, robotics," says Luckey. "And the companies that did have expertise, like Google, like Facebook, like Apple, were refusing to work with the U.S. national security community."

Andurils pitch is AI weapons, built in less time and at a lower cost than traditional defense contractors.

The man spreading this message is an iconoclastic figure in the largely liberal tech community for his work with the military, and his outspoken politics, including long-standing support for Donald Trump.

But Palmer Luckey is hard to ignore.

Just days after Russias full-scale invasion of Ukraine in 2022, Luckey made his way to the capital Kyiv, and met Ukrainian President Volodymyr Zelenskyy.

"Anduril has had hardware in Ukraine since the second week of the war. So we immediately got involved," Luckey says.

A Ukrainian soldier operates a drone in the Zaporizhzhia region of southeastern Ukraine on June 14. Ukraine has been using cheap, civilian drones that it buys off the Internet. But Russia has responded with electronic jamming, often rendering such drones useless. Ukraine is now making its own drones and is looking for autonomous systems that are difficult to detect and stop. Andriy Andriyenko/AP hide caption

The Ukraine war has become a laboratory for an array of high-tech systems.

Most striking is Elon Musk's Starlink satellite network that provides critical communications for Ukraines military.

However, in this emerging industry of AI weapons, critics say a lot of bugs still need to be worked out.

In several off-the-record conversations, people working closely with Ukraines military say many new weapons, from a range of companies, still have flaws, are vulnerable to Russian counter-measures, and simply have not yet performed as advertised.

So far, they add, these weapons have had a limited impact and have not changed the wars trajectory.

Andurils CEO Brian Schimpf acknowledges the difficulties, but sees them as surmountable.

Brian Schimpf, the CEO of Anduril, stands next to a a launch system for the Altius drone at company headquarters. Schimpf acknowledges that operating in Ukraine is a challenge, but says Anduril's AI weapons are designed to be updated quickly to adapt to changing conditions. Philip Cheung for NPR hide caption

"Ukraine is a very challenging environment to learn in," he says. "Ive heard various estimates from the Ukrainians themselves that any given drone typically has a life span of about four weeks. The question is can you respond and adapt?

Jacquelyn Schneider, who studies military technology as a fellow at the Hoover Institution, says the war has dramatically increased the pace of innovation.

"Technologies that worked really well even a few months ago are now constantly having to change," she says. "And the big difference I do see is that software changes the rate of change."

Weapons systems in Ukraine need to be updated frequently, just like the software on a phone or computer.

"If you're buying a weapons platform that cannot be very easily modified for these software innovations, then the weapon system will become useless or not as effective in a very short period of time," she adds.

P.W. Singer, an author who writes about war and tech, says, "There's this mythology of innovation as if it happens in one place."

The reality is "there's a lot of cool, exciting stuff happening in the big defense primes. There's a lot of cool, exciting stuff happening in the big-tech Silicon Valley companies. There's a lot of cool, exciting stuff happening in small startups," he says.

He also says AI weapons like drones should be seen as an addition not a replacement for existing weapons.

"No one is saying, 'Well, that means there's no need for our traditional military. There's no need for manned airplanes.' Of course, you need both," he says.

Anduril's Altius 600 drone hangs in the showroom at company headquarters in southern California. Most drones are remotely controlled by a pilot. The Anduril drones can be programmed before takeoff to search, find and strike a target without being guided by a remote pilot. Philip Cheung for NPR hide caption

In the face of Russia's big offensive two years ago, Ukrainians turned to small, cheap civilian drones made in China and available on the Internet. The Ukrainians attached grenades and other small explosives, then dropped the weapons down the open turrets of unsuspecting Russian tanks.

In many instances, a $1,000 drone was taking out a multi-million-dollar tank or other expensive Russian weaponry, as well as inflicting casualties on Russian troops. But it's getting much harder for the Ukrainians to carry out these kinds of attacks.

The Russians have responded with electronic jamming, blocking the signal the drone was sending to the Ukrainian soldier operating it. This renders the drone useless.

This is where Anduril is trying to step in. The company's AI drones can be programmed before takeoff to search on their own for Russian tanks or other targets.

Once launched, these drones dont need guidance from a Ukrainian soldier making them very hard to detect and stop, says Luckey.

"The autonomy onboard is really what sets it apart," he says. "It's not a remote controlled plane. There's a brain on it that is able to look for targets, identify targets and fly into those targets."

Of course, this raises questions about whos responsible if something goes wrong like hitting civilians.

In a recent report submitted to the United Nations, Human Rights Watch called for "the urgent negotiation and adoption of a legally binding instrument to prohibit and regulate autonomous weapons systems."

The organization says more than 270 groups and 70 countries have now joined its Campaign to Stop Killer Robots.

However, Andurils Brian Schimpf says AI weapons are "not about taking humans out of the loop. I don't think that's the right ethical framework. This is really about how do we make human decision-makers more effective and more accountable to their decisions."

Ukraine now makes its own sea drones essentially jet skis packed with explosives which have inflicted serious damage on the Russian navy in the Black Sea.

Luckey puts on a virtual reality headset to show an augmented reality model of Andurils Dive-LD underwater drone at Anduril headquarters. Philip Cheung for NPR hide caption

Luckey shows me Andurils version, an underwater drone called Dive-LD, in an old, largely vacant industrial building thats part of Andurils otherwise shiny campus.

We put on virtual reality headsets an updated model of the one Luckey created for an augmented look at the sub.

"It's an autonomous underwater vehicle that is able to go very, very long distances, dive to a depth of about 6,000 meters, which is deep enough to go to the bottom of almost any ocean," he says.

Last month, Anduril won a U.S. Navy contract to build more than 200 of them annually.

Luckey has pursued his interests in tech, business and politics since his teen years. Way back in 2011, Luckey wrote to Donald Trump and urged him to run for president.

"I said, 'Hey, consider me one of the people who thinks it's good to have a businessperson in office, somebody who's familiar with signing both sides of a check.'"

He still supports Trump today.

"In general, yeah, I think he'd make a good commander in chief," he adds.

Yet from a business perspective, he says hes not that concerned about who wins in November.

"We made a lot of money under Trump. We made even more money under Biden. I think we're going to continue expanding whoever is in office next," said Luckey.

More AI weapons are coming, he says, no matter whos in the White House.

Greg Myre is an NPR national security correspondent. Follow him @gregmyre1.

NPR producer Kira Wakeam contributed to this report.

Read the rest here:

He created Oculus headsets as a teenager. Now he makes AI weapons for Ukraine - NPR