Report: Some AI Systems Used Images of Child Sexual Abuse

 An illustration shows the words "Artificial Intelligence" and letters and a robot hand miniature, June 23, 2023. (Reuters Photo)
An illustration shows the words "Artificial Intelligence" and letters and a robot hand miniature, June 23, 2023. (Reuters Photo)
TT

Report: Some AI Systems Used Images of Child Sexual Abuse

 An illustration shows the words "Artificial Intelligence" and letters and a robot hand miniature, June 23, 2023. (Reuters Photo)
An illustration shows the words "Artificial Intelligence" and letters and a robot hand miniature, June 23, 2023. (Reuters Photo)

Some artificial intelligence (AI) image-generators used thousands of images of child sexual abuse without the parents’ consent, according to a new report, The Associated Press reported.

The report revealed that the images have made it easier for AI systems to produce realistic and explicit imagery of fake children as well as transform social media photos of fully clothed real teens into nudes, much to the alarm of schools and law enforcement around the world.

Until recently, anti-abuse researchers thought the only way that some unchecked AI tools produced abusive imagery of children was by essentially combining what they’ve learned from two separate buckets of online images — adult pornography and benign photos of kids.

But the Stanford Internet Observatory found more than 3,200 images of suspected child sexual abuse in the giant AI database LAION, an index of online images and captions that’s been used to train leading AI image-makers such as Stable Diffusion.

The watchdog group based at Stanford University worked with the Canadian Center for Child Protection and other anti-abuse charities to identify the illegal material and report the original photo links to law enforcement.

On the eve of the release of the Stanford Internet Observatory’s report, LAION told The Associated Press it was temporarily removing its datasets.

LAION said in a statement that it “has a zero-tolerance policy for illegal content and in an abundance of caution, we have taken down the LAION datasets to ensure they are safe before republishing them.”

The Stanford report stressed that any photos of children — even the most benign — should not be fed into AI systems without their family’s consent.



Major Meta Trial Begins as Lawyers Spar over Witnesses, Damages

The Meta company logo at the entrance of its temporary pavilion before the World Economic Forum in Davos, Switzerland, January 18, 2025 (Reuters)
The Meta company logo at the entrance of its temporary pavilion before the World Economic Forum in Davos, Switzerland, January 18, 2025 (Reuters)
TT

Major Meta Trial Begins as Lawyers Spar over Witnesses, Damages

The Meta company logo at the entrance of its temporary pavilion before the World Economic Forum in Davos, Switzerland, January 18, 2025 (Reuters)
The Meta company logo at the entrance of its temporary pavilion before the World Economic Forum in Davos, Switzerland, January 18, 2025 (Reuters)

Facebook-parent Meta will defend itself in a landmark social media trial that begins Tuesday in California, where attorneys have sparred over witnesses and financial penalties.

A coalition of states sued Meta in 2023, and the four states now going to trial were selected to represent them.

Lawyers for the states will argue Meta deliberately made Facebook and Instagram addictive for children, in violation of state and federal laws, reported AFP.

In recent days Meta, which has more than three billion users around the world, tried to block ex-Meta employee and expert witness, Arturo Bejar, from testifying.

Federal Judge Yvonne Gonzalez Rogers dismissed the company's request, handing the states an early victory. In her ruling, she described it as a "Hail Mary" effort to "eliminate a strong witness" for the plaintiffs.

Bejar has previously testified against the company, including in a New Mexico trial which Meta lost.

Lawyers for California, Colorado, Kentucky and New Jersey are expected to question Bejar about Meta's practices around safety and growth, and whether the company publicly misrepresented what it knew, according to a court document.

In a new filing Monday, Meta asked the judge to restrict the scope of testimony from another potential expert witness, Colin Gray, whom the states intend to ask about "dark patterns" -- features designed to manipulate users into making choices preferred by a company.

Meta founder and chief Mark Zuckerberg is among the star witnesses expected to testify.

Meta "strongly disagrees" with the allegations in the trial, a spokesperson told AFP.

During a hearing last week, a lawyer for the states said they are seeking around $200 billion -- not penalties exceeding $1 trillion, as Meta had claimed in a court filing.

The lawyer said they believe Meta calculated that figure "for shock value."

In addition to financial penalties, the states are demanding changes to Meta's apps.

Eight people were selected last week to serve on an advisory jury, but the judge will make the final decision in the case. The trial is expected to last six weeks and a verdict is expected by early October.

Although it's not the first lawsuit that seeks to hold a social media company accountable for mental health and safety issues, it could become one of the most consequential.

- 'Broader reckoning' -

Last week, Rogers and lawyers for both sides probed prospective jurors about their views of social media and Meta, asking if they or their children had social media accounts and whether social media is responsible for mental health, among other questions.

One prospective juror said they believed social media plays a role in mental health and compared the serotonin boost that some people feel from "doomscrolling" to using cocaine.

The case could be "the beginning of a broader reckoning" for Meta, Nora Freeman Engstrom told AFP via email. She is a law professor and associate dean at Stanford.

It will be important to see the "gap" between what Meta knew privately and what it disclosed publicly, Engstrom said.

"The huge issue here is reputational harm" and being forced to make major changes, Vincent Joralemon, a director at Berkeley's Life Sciences Law and Policy Center, told AFP.

Experts see parallels with a three-decade-old settlement between dozens of US states and tobacco companies.

"It really feels like tobacco in the 1990s," Joralemon said.

While cases about social media harms revolve around the intersection of technology and addiction, the case against Meta focuses on its business practices, similar to when US regulators sued tobacco companies, Joralemon said.

Dozens of US states sued four major tobacco companies for downplaying the harmful health impacts of their products, and won a 1998 landmark settlement that included financial penalties and changes to product marketing.

Those tobacco companies have paid over $176 billion since, according to data from the National Association of Attorneys General.

They will continue to pay $9 billion annually, according to the settlement.


China’s Unitree Unveils ‘Superman’ Robot as Fervor Builds Ahead of Shanghai Debut

Children stand near Unitree's humanoid robots on display at a Unitree store in Beijing, China August 10, 2026. (Reuters)
Children stand near Unitree's humanoid robots on display at a Unitree store in Beijing, China August 10, 2026. (Reuters)
TT

China’s Unitree Unveils ‘Superman’ Robot as Fervor Builds Ahead of Shanghai Debut

Children stand near Unitree's humanoid robots on display at a Unitree store in Beijing, China August 10, 2026. (Reuters)
Children stand near Unitree's humanoid robots on display at a Unitree store in Beijing, China August 10, 2026. (Reuters)

China's Unitree unveiled a new high-speed robot it dubbed "Superman" on Monday, days before becoming the first general-purpose robotics company to list on mainland China's stock market.

Unitree, the world's biggest humanoid-robot maker by sales, will start trading in Shanghai on Wednesday after raising 6.1 billion yuan ($905 ‌million) in ‌an initial public offering last week.

The company, based ‌in ⁠the eastern Chinese ⁠tech hub of Hangzhou, said it developed the new robot in just over three months. It achieved a standing jump of 2 meters and reached a top speed of 12.66 meters per second.

The Shanghai debut coincides with the opening of the World Robot Conference in Beijing, adding to investor attention ⁠on a company whose humanoid robots have drawn ‌global notice for running, dancing and ‌performing martial arts.

The IPO was more than 8,000 times oversubscribed ‌by retail investors, a record for Shanghai's tech-focused STAR Market.

In ‌a social media post last week, Unitree said it had cumulatively produced and delivered about 18,000 bipedal humanoid robots across multiple models as of July.

DERIVATIVES MARKET POINTS TO STRONG DEBUT

Unitree said the robot ‌remained a work in progress and could be improved further in coming months.

Founder Wang Xingxing ⁠gained national ⁠prominence after securing a front-row seat at a summit hosted by Chinese President Xi Jinping for leading technology entrepreneurs last year.

The startup is also backed by some of China's most influential technology companies including Tencent, Alibaba, and DeepSeek.

Ahead of Unitree's Shanghai listing, derivative trading in offshore markets has pointed to a strong debut.

Its shares may also benefit from a broader recovery in Chinese technology stocks following last month's selloff.

The STAR50 Index has rebounded about 15% from a three-month low hit in early August.


Nvidia to Invest $1.5 billion in SB Energy under OpenAI Data Center Deal

FILE PHOTO: Nvidia logo is seen in this illustration taken June 11, 2026. REUTERS/Dado Ruvic/Illustration/File Photo
FILE PHOTO: Nvidia logo is seen in this illustration taken June 11, 2026. REUTERS/Dado Ruvic/Illustration/File Photo
TT

Nvidia to Invest $1.5 billion in SB Energy under OpenAI Data Center Deal

FILE PHOTO: Nvidia logo is seen in this illustration taken June 11, 2026. REUTERS/Dado Ruvic/Illustration/File Photo
FILE PHOTO: Nvidia logo is seen in this illustration taken June 11, 2026. REUTERS/Dado Ruvic/Illustration/File Photo

Nvidia will invest $1.5 billion in SoftBank-backed SB Energy and secure up to 8 gigawatts of AI computing capacity at an Ohio campus being built by the data center developer for OpenAI.

The deal is the latest where Nvidia is financing the ecosystem consuming its chips, a strategy that has helped fuel demand but also drawn scrutiny over the circular flows of funds from the chipmaker to its biggest customers, Reuters reported.

Leading tech firms are increasingly tying together chips, power and data center development as they race to secure the infrastructure needed for increasingly power-hungry AI models.

Chip giant Nvidia has secured land and power at Ohio's PORTS-Pike Technology Campus for an AI data center that will use its graphics processors and networking gear, with an initial capacity of 4.25 GW.

SB Energy and SoftBank plan to build at least 10 GW of new power generation and invest $4.2 billion in Ohio grid infrastructure to support AI data centers.

Also backed by OpenAI, SB Energy develops large-scale power and data center infrastructure projects. Founded in 2019, the company is building several data center campuses to support rising demand tied to AI workloads.