PARIS — When French prosecutors took aim at Telegram boss Pavel Durov, they had a trump card to wield – a tough new law with no international equivalent that criminalizes tech titans whose platforms allow illegal products or activities.

The so-called LOPMI law, enacted in January 2023, has placed France at the forefront of a group of nations taking a sterner stance on crime-ridden websites. But the law is so recent that prosecutors have yet to secure a conviction.

With the law still untested in court, France’s pioneering push to prosecute figures like Durov could backfire if its judges balk at penalizing tech bosses for alleged criminality on their platforms.

A French judge placed Durov under formal investigation last month, charging him with various crimes, including the 2023 offence: “Complicity in the administration of an online platform to allow an illicit transaction, in an organized gang,” which carries a maximum 10-year sentence and a $556,300 fine.

Being under formal investigation does not imply guilt or necessarily lead to trial, but indicates judges think there’s enough evidence to proceed with the probe. Investigations can last years before being sent to trial or dropped.

Durov, out on bail, denies Telegram was an “anarchic paradise.” Telegram has said it “abides by EU laws,” and that it’s “absurd to claim that a platform or its owner are responsible for abuse of that platform.”

In a radio interview last week, Paris Prosecutor Laure Beccuau hailed the 2023 law as a powerful tool for battling organized crime groups who are increasingly operating online.

The law appears to be unique. Eight lawyers and academics told Reuters they were unaware of any other country with a similar statute.

“There is no crime in U.S. law directly analogous to that and none that I’m aware of in the Western world,” said Adam Hickey, a former U.S. deputy assistant attorney general who established the Justice Department’s (DOJ) national security cyber program.

Hickey, now at U.S. law firm Mayer Brown, said U.S. prosecutors could charge a tech boss as a “co-conspirator or an aider and abettor of the crimes committed by users” but only if there was evidence the “operator intends that its users engage in, and himself facilitates, criminal activities.”

He cited the 2015 conviction of Ross Ulbricht, whose Silk Road website hosted drug sales. U.S. prosecutors argued Ulbricht “deliberately operated Silk Road as an online criminal marketplace … outside the reach of law enforcement,” according to the DOJ. Ulbricht got a life sentence.

Timothy Howard, a former U.S. federal prosecutor who put Ulbricht behind bars, was “skeptical” Durov could be convicted in the United States without proof he knew about the crimes on Telegram, and actively facilitated them – especially given Telegram’s vast, mainly law-abiding user base.

“Coming from my experience of the U.S. legal system,” he said, the French law appears “an aggressive theory.”

Michel Séjean, a French professor of cyber law, said the toughened legislation in France came after authorities grew exasperated with companies like Telegram.

“It’s not a nuclear weapon,” he said. “It’s a weapon to prevent you from being impotent when faced with platforms that don’t cooperate.”

Tougher laws

The 2023 law traces its origins to a 2020 French interior ministry white paper, which called for major investment in technology to tackle growing cyber threats.

It was followed by a similar law in November 2023, which included a measure for the real-time geolocation of people suspected of serious crimes by remotely activating their devices. A proposal to turn on their devices’ cameras and mouthpieces so that investigators could watch or listen in was shot down by France’s Constitutional Council.

These new laws have given France some of the world’s toughest tools for tackling cybercrime, with the proof being the arrest of Durov on French soil, said Sadry Porlon, a French lawyer specialized in communication technology law.

Tom Holt, a cybercrime professor at Michigan State University, said LOPMI “is a potentially powerful and effective tool if used properly,” particularly in probes into child sexual abuse images, credit card trafficking and distributed denial of service attacks, which target businesses or governments.

Armed with fresh legislative powers, the ambitious J3 cybercrime unit at the Paris prosecutor’s office, which is overseeing the Durov probe, is now involved in some of France’s most high-profile cases.

In June, the J3 unit shut down Coco, an anonymized chat forum cited in over 23,000 legal proceedings since 2021 for crimes including prostitution, rape and homicide.

Coco played a central role in a current trial that has shocked France.

Dominique Pelicot, 71, is accused of recruiting dozens of men on Coco to rape his wife, whom he had knocked out with drugs. Pelicot, who is expected to testify this week, has admitted his guilt, while 50 other men are on trial for rape.

Coco’s owner, Isaac Steidel, is suspected of a similar crime as Durov: “Provision of an online platform to allow an illicit transaction by an organized gang.”

Steidel’s lawyer, Julien Zanatta, declined to comment.
…

WASHINGTON — “I love you, China. My dear mother,” former U.S. President Donald Trump, standing in front of a mic at a lectern, appears to sing in perfect Mandarin.

“I cry for you, and I also feel proud for you,” Vice President Kamala Harris, Trump’s Democratic opponent in this year’s election, appears to respond, also in perfect Mandarin. Trump lets out a smile as he listens to the lyric.

The video has received thousands of likes and tens of thousands of reposts on Douyin, China’s variation of TikTok.

“These two are almost as Chinese as it gets,” one comment says.

Neither Trump nor Harris knows Mandarin. And the duet shown in the video has never happened. But recently, deepfake videos, frequently featuring top U.S. leaders, including President Joe Biden, singing Chinese pop songs, have gone viral on the Chinese internet.

Some of the videos have found their way to social media platforms not available in China, such as Instagram, TikTok and X.

U.S. intelligence officials and experts have long warned about how China and other foreign adversaries have been implementing generative AI in their disinformation effort to disrupt and influence the 2024 presidential election.

“There has been an increased use of Chinese AI-generated content in recent months, attempting to influence and sow division in the U.S. and elsewhere,” a Microsoft report on China’s disinformation threat said in April.

Few of the people who saw the videos of the American leaders singing in Chinese, however, were convinced that they were real, based on what users wrote in the comments. The videos themselves do not contain misinformation, either.

Instead, these videos and their popularity reflect, at least in part, a sense of cultural confidence in Chinese netizens in the age of perpetually intensifying U.S.-China competitions, observers told VOA Mandarin.

By making the likes of Biden and Trump sing whatever Chinese songs the creators of the videos want them to sing, they can “culturally domesticate powerful Americans,” said Alexa Pan, a researcher on China’s AI industry for ChinaTalk, an influential newsletter about China and technology.

“Making fun of U.S. leaders might be especially politically acceptable to and popular with Chinese viewers,” she said.

Political opponents sing about friendship

Videos of American leaders singing in Chinese started to spread on Chinese social media in May. In many of the videos featuring Biden and Trump, creators made the two politically opposed men sing songs about friendship.

After Biden announced his withdrawal from the presidential race in July, one viral video had him sing to Trump, “Actually I don’t want to leave. Actually, I want to stay. I want to stay with you through every spring, summer, autumn and winter,” to which Trump appeared to sing, “You have to believe me. It won’t take long before we can spend our whole life together.”

“Crying eyes,” one Chinese netizen commented sarcastically. “They must have gotten along really well.”

Another such video posted on Instagram received mostly positive reactions. Some users said it was a stark contrast to the bitterness that has permeated U.S. politics.

“Made me laugh,” an Instagram user wrote. “Wouldn’t that be so refreshing to actually have them sing like that together?”

Easy to make

After reviewing some of the videos, Pan, of ChinaTalk, told VOA Mandarin that she believes they were quite easy to make.

Obvious flaws in the videos, including body parts occasionally blending into the background, suggest they were created with simple AI technology, Pan said.

“One could generate these videos on the many AI text-to-video generation platforms available in China,” she wrote in an e-mail.

On the Chinese internet, there are countless tutorials on how to make AI-generated videos using popular lip-syncing AI models, such as MuseTalk, released by Chinese tech giant Tencent, and SadTalker, developed by Xi’an Jiaotong University, a research-focused university in northwestern China.

One Douyin account reviewed by VOA Mandarin has pumped out over 200 videos of American leaders singing in Chinese since May. One of the account’s videos was even reposted by the Iranian embassy.

Chinese leaders off-limits

The release of ChatGPT by OpenAI in 2022 has triggered a global AI frenzy, with China being one of the leading countries developing the technology. The United Nations said in July that China had requested the most patents on generative AI, with the U.S. being a distant second.

On the Chinese internet, the obsession has been particularly strong with deepfakes, which can be used to manipulate videos, images and audio of people to make them appear to say or sing things that they have not actually uttered.

Some deepfake videos are made mostly for fun, such is the case with Biden and Trump singing Chinese songs. But there have also been abuses of the technology. Earlier this year, web users in China stole a Ukrainian girl’s image and turned her into a “Russian beauty” to sell goods online.

 China has released strict regulations on deepfakes. A 2022 law states that the technology cannot be used to “endanger the national security and interests, harm the image of the nation, harm the societal public interest, disturb economic or social order, or harm the lawful rights and interests of others.”

Yang Han, an Australian commentator who used to work for China’s Foreign Ministry, told VOA Mandarin that the prominence of U.S leaders and the absence of Chinese leaders in these viral AI videos reflects a lack of political free speech in China.

He said that it reminds him of a joke that former U.S. President Ronald Reagan used to tell during the Cold War.

“An American and a Russian compare with each other whose country has more freedom,” Yang said, relaying the joke. “The American says he can stand in front of the White House and call Reagan stupid. The Russian dismisses it and says he can also stand in front of the Kremlin and call Reagan stupid.”
…

…

…

tokyo — A long robot entered a damaged reactor at Japan’s Fukushima nuclear power plant on Tuesday, beginning a two-week, high-stakes mission to retrieve for the first time a tiny amount of melted fuel debris from the bottom.

The robot’s trip into the Unit 2 reactor is a crucial initial step for what comes next — a daunting, decades-long process to decommission the plant and deal with large amounts of highly radioactive melted fuel inside three reactors that were damaged by a massive earthquake and tsunami in 2011. Specialists hope the robot will help them learn more about the status of the cores and the fuel debris.

Here is an explanation of how the robot works, its mission, significance and what lies ahead as the most challenging phase of the reactor cleanup begins.

What is the fuel debris?

Nuclear fuel in the reactor cores melted after the magnitude 9.0 earthquake and tsunami in March 2011 caused the Fukushima Daiichi nuclear plant’s cooling systems to fail. The melted fuel dripped down from the cores and mixed with internal reactor materials such as zirconium, stainless steel, electrical cables, broken grates and concrete around the supporting structure and at the bottom of the primary containment vessels.

The reactor meltdowns caused the highly radioactive, lava-like material to spatter in all directions, greatly complicating the cleanup. The condition of the debris also differs in each reactor.

Tokyo Electric Power Company Holdings, or TEPCO, which manages the plant, says an estimated 880 tons of molten fuel debris remains in the three reactors, but some experts say the amount could be larger.

What is the robot’s mission?

Workers will use five 1.5-meter-long pipes connected in sequence to maneuver the robot through an entry point in the Unit 2 reactor’s primary containment vessel. The robot itself can extend about 6 meters inside the vessel. Once inside, it will be maneuvered remotely by operators at another building at the plant because of the fatally high radiation emitted by the melted debris.

The front of the robot, equipped with tongs, a light and a camera, will be lowered by a cable to a mound of melted fuel debris. It will then snip off and collect a bit of the debris — less than 3 grams). The small amount is meant to minimize radiation dangers.

The robot will then back out to the place it entered the reactor, a roundtrip journey that will take about two weeks.

The mission takes that long because the robot must make extremely precise maneuvers to avoid hitting obstacles or getting stuck in passageways. That has happened to earlier robots.

TEPCO is also limiting daily operations to two hours to minimize the radiation risk for workers in the reactor building. Eight six-member teams will take turns, with each group allowed to stay maximum of about 15 minutes.

What do officials hope to learn?

Sampling the melted fuel debris is “an important first step,” said Lake Barrett, who led the cleanup after the 1979 disaster at the U.S. Three Mile Island nuclear plant for the Nuclear Regulatory Commission and is now a paid adviser for TEPCO’s Fukushima decommissioning.

While the melted fuel debris has been kept cool and has stabilized, the aging of the reactors poses potential safety risks, and the melted fuel needs to be removed and relocated to a safer place for long-term storage as soon as possible, experts say.

An understanding of the melted fuel debris is essential to determine how best to remove it, store it and dispose of it, according to the Japan Atomic Energy Agency.

Experts expect the sample will also provide more clues about how exactly the meltdown 13 years ago played out, some of which is still a mystery.

The melted fuel sample will be kept in secure canisters and sent to multiple laboratories for more detailed analysis. If the radiation level exceeds a set limit, the robot will take the sample back into the reactor.

“It’s the start of a process. It’s a long, long road ahead,” Barrett said in an online interview. “The goal is to remove the highly radioactive material, put it into engineered canisters … and put those in storage.”

For this mission, the robot’s small tong can only reach the upper surface of the debris. The pace of the work is expected to pick up in the future as more experience is gained and robots with additional capabilities are developed.

What’s next?

TEPCO will have to “probe down into the debris pile, which is over a meter thick, so you have to go down and see what’s inside,” Barrett said, noting that at Three Mile Island, the debris on the surface was very different from the material deeper inside. He said multiple samples from different locations must be collected and analyzed to better understand the melted debris and develop necessary equipment, such as stronger robots for future larger-scale removal.

Compared to collecting a tiny sample for analysis, it will be a more difficult challenge to develop and operate robots that can cut larger chunks of melted debris into pieces and put that material into canisters for safe storage.

There are also two other damaged reactors, Unit 1 and Unit 3, which are in worse condition and will take even longer to deal with. TEPCO plans to deploy a set of small drones in Unit 1 for a probe later this year and is developing even smaller “micro” drones for Unit 3, which is filled with a larger amount of water.

Separately, hundreds of spent fuel rods remain in unenclosed cooling pools on the top floor of both Unit 1 and 2. This is a potential safety risk if there’s another major quake. Removal of spent fuel rods has been completed at Unit 3.

When will the decommissioning be finished?

Removal of the melted fuel was initially planned to start in late 2021 but has been delayed by technical issues, underscoring the difficulty of the process. The government says decommissioning is expected to take 30-40 years, while some experts say it could take as long as 100 years.

Others are pushing for an entombment of the plant, as at Chernobyl after its 1986 explosion, to reduce radiation levels and risks for plant workers.

That won’t work at the seaside Fukushima plant, Barrett says.

“You’re in a high seismic area, you’re in a high-water area, and there are a lot of unknowns in those (reactor) buildings,” he said. “I don’t think you can just entomb it and wait.”
…

…