Amazon.com Inc. reported hundreds of thousands of pieces of content last year that it believed included child sexual abuse, which it found in data gathered to improve its artificial intelligence models. Though Amazon removed the content before training its models, child safety officials said the company has not provided information about its source, potentially hindering law enforcement from finding perpetrators and protecting victims. 

Throughout last year, Amazon detected the material in its AI training data and reported it to the National Center for Missing and Exploited Children, or NCMEC. The organization, which was established by Congress to field tips about child sexual abuse and share them with law enforcement, recently started tracking the number of reports specifically tied to AI products and their development. In 2025, NCMEC saw at least a fifteen-fold increase in these AI-related reports, with “the vast majority” coming from Amazon. The findings haven’t been previously reported.

An Amazon spokesperson said the training data was obtained from external sources, and the company doesn’t have the details about its origin that could aid investigators. It’s common for companies to use data scraped from publicly available sources, such as the open web, to train their AI models. Other large tech companies have also scanned their training data and reported potentially exploitative material to NCMEC. However, the clearinghouse pointed to “glaring differences” between Amazon and its peers. The other companies collectively made just “a handful of reports,” and provided more detail on the origin of the material, a top NCMEC official said.

In an emailed statement, the Amazon spokesperson said that the company is committed to preventing child sexual abuse material across all of its businesses. “We take a deliberately cautious approach to scanning foundation model training data, including data from the public web, to identify and remove known [child sexual abuse material] and protect our customers,” the spokesperson said. 

The spike in Amazon’s reports coincides with a fast moving AI race that has left companies large and small scrambling to acquire and ingest huge volumes of data to improve their models. But that race has also complicated the work of child safety officials — who are struggling to keep up with the changing technology — and challenged regulators tasked with safeguarding AI from abuse. AI safety experts warn that quickly amassing large datasets without proper safeguards comes with grave risks. 

Amazon accounted for most of the more than 1 million AI-related reports of child sexual abuse material submitted to NCMEC in 2025, the organization said. It marks a jump from the 67,000 AI-related reports that came from across the tech and media industry a year prior, and just 4,700 in 2023. This category of AI-related reports can include AI-generated photos and videos, or sexually explicit conversations with AI chatbots. It can also include photos of real victims of sexual abuse that were collected, even unintentionally, in an effort to improve AI models. 

Training AI on illegal and exploitative content raises newfound concerns. It could risk shaping a model’s underlying behaviors, potentially improving its ability to digitally alter and sexualize photos of real children or create entirely new images of sexualized children that never existed. It also raises the threat of continuing the circulation of the images that models were trained on — re-victimizing children who have suffered abuse. 

The Amazon spokesperson said that, as of January, the company is “not aware of any instances” of its models generating child sexual abuse material.  None of its reports submitted to NCMEC were of AI-generated material, the spokesperson added. Instead, the content was flagged by an automatic detection tool that compared it against a database of known child abuse material involving real victims, a process called “hashing.” Approximately 99.97% of the reports resulted from scanning “non-proprietary training data,” the spokesperson said. 

Amazon believes it over-reported these cases to NCMEC to avoid accidentally missing something. “We intentionally use an over-inclusive threshold for scanning, which yields a high percentage of false positives,” the spokesperson added.

Amazon has more than 900 data center facilities worldwide.
Photo Credit: Bloomberg

The AI-related reports received last year are just a fraction of the total number submitted to NCMEC. The larger category of reports also includes suspected child sexual abuse material sent in private messages or uploaded to social media feeds and the cloud. In 2024, for example, NCMEC received more than 20 million reports from across industry, with most coming from Meta Platforms Inc. subsidiaries Facebook, Instagram and WhatsApp. Not all reports are ultimately confirmed as containing child sexual abuse material, referred to with the acronym CSAM.

Still, the volume of suspected CSAM that Amazon detected across its AI pipeline in 2025 stunned child safety experts interviewed by Bloomberg News. The hundreds of thousands of reports made to NCMEC marked a drastic surge for the company. In 2024, Amazon and all of its subsidiaries made a total of 64,195 reports. 

“This is really an outlier,” said Fallon McNulty, the executive director of NCMEC’s CyberTipline, the entity to which US-based social media platforms, cloud providers and other companies are legally required to report suspected CSAM. “Having such a high volume come in throughout the year begs a lot of questions about where the data is coming from, and what safeguards have been put in place.” 

McNulty, speaking in an interview, said she has little visibility into what’s driving the surge of sexually exploitative material in Amazon’s initial training data sets. Amazon has provided “very little to almost no information” in their reports about where the illicit material originally came from, who had shared it, or if it remains actively available on the internet, she said. 

While Amazon is not required to share this level of detail, the lack of information makes it impossible for NCMEC to track down the material’s origin and work to get it removed, McNulty said. It also limits relevant law enforcement agencies tasked with searching for sex offenders and children in active danger. “There’s nothing then that can be done with those reports,” she said. “Our team has been really clear with [Amazon] that those reports are inactionable.”

When asked why the company didn’t disclose information about the possible origin of the material, or other key details, the Amazon spokesperson replied, “because of how this data is sourced, we don’t have the data that comprises an actionable report.” The spokesperson did not explain how the third-party data was sourced or why the company did not have sufficient information to create actionable reports. “While our proactive safeguards cannot provide the same detail in NCMEC reports as consumer-facing tools, we stand by our commitment to responsible AI and will continue our work to prevent CSAM,” the spokesperson said.

NCMEC, a nonprofit, receives funding both from the US government and private industry. Amazon is among its funders and holds a corporate seat on its board. 

“There should be more transparency on how companies are gathering and analyzing the data to train their models — and how they’re training them,” said David Thiel, the former chief technologist at the Stanford Internet Observatory, who has researched the prevalence of child sexual abuse material in AI training data. 

Such data can be licensed, purchased or scraped from the internet, or could be so-called synthetic data, which is text or images created by other AI tools. As AI companies seek to release new models quickly, “the rapid gathering of data is a much higher priority than doing safety analyses,” Thiel said. He warned that there are “always some errors” when it comes to sifting out CSAM from training data, and believes the industry needs to be more open about where its data is coming from. 

Amazon’s Bedrock offering, which gives customers access to various AI models so they can build their own AI products, includes automated detection for known CSAM and rejects and reports positive matches. The company’s consumer-facing generative AI products also allow users to report content that escapes its controls. 

The Seattle-based tech giant scans for CSAM across its other businesses, too, including its consumer photo storage service. Amazon’s cloud computing division, Amazon Web Services, also removes CSAM when it’s discovered on the web services it hosts. McNulty said AWS submitted far fewer reports than came from Amazon’s AI efforts. Amazon declined to break out specific reporting data across its various business units, but noted it would share broad data in March.

Only recently have technology companies really begun to scrutinize their AI models and training data for CSAM, said David Rust-Smith, a data scientist at Thorn, a nonprofit organization that provides tools to companies, including Amazon, to detect the exploitative material. 

“There’s definitely been a big shift in the last year of people coming to us asking for help cleaning data sets,” said Rust-Smith. He noted that “some of the biggest players” have sought to apply Thorn’s detection tools to their training data, but declined to speak about any individual company. Amazon did not use Thorn’s technology to scan its training data, the spokesperson confirmed.

Rust-Smith said AI-focused companies are approaching Thorn with a newfound urgency. “People are learning what we already knew, which is, if you hoover up a ton of the internet, you’re going to get [child sexual abuse material],” he said. 

Amazon was not the only company to spot and report potential CSAM from its AI workflows last year. Alphabet Inc.’s Google and OpenAI told Bloomberg News that they scan AI training data for exploitative material — a process that has surfaced potential CSAM, which the  companies then reported to NCMEC. Meta and Anthropic PBC said they, too, search training data for CSAM. Meta did not comment on whether it had identified the material, but said it would report to NCMEC if it did. Anthropic said it has not reported such material out of its training data. Meta and Google said that they’ve taken efforts to ensure that reports related to their AI workflows are distinguishable from those generated by others parts of their business. 

McNulty said that, with the exception of Amazon, the AI-related reports it received last year came in “really, really small volumes,” and included key details that allowed the clearinghouse to pass on actionable information to law enforcement. 

“Simply flagging that you came across something but not providing any type of actionable detail doesn’t help the larger child safety space,” McNulty said.

Sydney:

Former childcare worker Ashley Paul Griffith on Monday pleaded guilty to raping, sexually abusing and exploiting dozens of girls under his care in Australia and Italy.

It took Judge Anthony Rafter more than two hours to read out the 307 charges against Griffith in the Brisbane courtroom where several victims and their families had gathered, according to state broadcast ABC.

Police accused the 46-year-old of being one of Australia’s worst-ever paedophiles after the charges against him were made public last year.

Griffith was first arrested in 2022 for making child sexual abuse content. A year later he was charged with 1,623 offences against 91 children.

However, some charges were dropped and the ABC reported that Monday’s charges related to roughly 60 children. Many of the victims were under the age of 12.

The offences happened between 2003 and 2022 in twelve different locations across Australia as well as Pisa, Italy.

Griffith is in custody and will be sentenced at a later date. The court heard sentencing could take more than two days to give time to read victim impact statements, the ABC reported.

(Except for the headline, this story has not been edited by NDTV staff and is published from a syndicated feed.)

Two women in China have been accused of child abuse after they separated a crying child from her grandmother and locked her in the toilet of a plane to “educate her”. According to the BBC, the incident took place on August 24 onboard a Juneyao Airlines flight from the southwestern city of Guiyang to Shanghai. The little girl, who was travelling with her grandmother, started crying during the flight, following which the two women passengers locked the distressed child in the toilet, saying that they trying to help others. 

The incident went viral on the Chinese internet after one of the two women, identified as Gou Tingting, posted a video of herself carrying the girl inside the toilet. The clip also showed the other woman telling the child that she could only leave the bathroom if she stopped crying. They presented themselves as trying to help others on board, per the BBC. However, many people on the internet accused the two of “bullying” the child and lacking empathy. 

The airline, on the other hand, said that the girl’s grandmother had agreed to let two women take the girl to the toilet. It also added that it had also spoken to the child’s mother, who was not on the flight, and who “expressed her understanding” of the women’s behaviour, The Guardian reported. 

But shortly after the video was posted online, many criticised the women for lacking “empathy” and “bullying” the child. Responding to the criticism, Ms Gou said she “prefers to take action rather than be a bystander”. “I just wanted to calm the child down and let everyone rest,” she wrote on Chinese social media platform Douyin. The woman also explained that some passengers had “moved to the back of the plane to escape the noise” while others stuffed tissue papers into their ears. 

Also Read | “Smell Is Horrendous”: Plane Passenger Appalled As Couple Changes Baby’s “Poopy Diaper” At Seat

However, Ms Gou’s explanation did little to quell the backlash, and her social media account has since been set to private. “Children cannot control their emotions when they are one or two years old. What’s wrong with crying? Didn’t you cry when you were young too?” one user wrote, per the BBC. 

Another user, concerned about the psychological impact on the girl, said, “We should be thinking about how public spaces can better accept and accommodate young children.”

But there were some who also defended the women, saying that their actions were justified as the girl’s grandmother had given her consent. “To be honest, some children cannot do without some education,” one user wrote.

A two-year-old girl died on Tuesday after spending several hours alone in a hot car. Parker Scholtes was the victim’s identity, as given by the Marana Police Department. Parker’s father, Christopher Scholtes, 37, reportedly told authorities that she slept out in the back seat of the car when she got home.

Temperatures in Marana soared to 109 degrees Fahrenheit (42.7 degrees Celsius) that day.

Emergency responders were called to the Scholtes residence on Tuesday afternoon after Parker was found unresponsive in the vehicle. Despite medical intervention, the child was pronounced dead at the scene.

According to a release by the police department, “during the initial investigation, the father, identified as 37-year-old Christopher Scholtes, told officers he arrived home, and the 2-year-old girl was still asleep in the vehicle in her car seat. He left the vehicle on and was running in the driveway, with the air conditioning on and the child inside. Christopher told detectives he did not want to wake his daughter up so she could continue sleeping in the vehicle. Christopher then went inside the residence. The mother was not home, but shortly after she arrived, the child was discovered in the vehicle, which was no longer running, and the air conditioning was off. 9-1-1 was called, and the mother performed CPR until NWFD and MPD arrived. NWFD performed life-saving measures, and the child was transported to Banner University Medical Center. Unfortunately, the child was pronounced dead.”

Marana Police Detectives were granted a search warrant the night of the incident and conducted a thorough investigation. During the investigation, detectives obtained video surveillance from nearby residences and determined that the child had been in the vehicle for approximately three hours.

On July 12 at 8:38 AM, Christopher Scholtes was arrested and booked into the Pima County Adult Detention Centre for 2nd Degree Murder and Child Abuse related to this tragic event.

A 63-year-old man in Singapore is under investigation for allegedly confining his 15-year-old daughter to a tiny market stall for nearly a year. According to the South China Morning Post (SCMP), the man, surnamed Tan, had been selling vegetables at a market on Circuit Road for a decade. He allegedly made his teen daughter live in one of the three stalls he rented at the wet market for 11 months. The 15-year-old was discovered when a neighbouring stall owner smelled excrement coming from Mr Tan’s stall and reported it to Singapore’s National Environment Agency. 

The staff from the agency discovered the girl in April, SCMP reported. The 15-year-old had been living in the six-square-metre space. Crammed into the small area was a desk, fridge, fan and a makeshift dirty bed on the floor. 

Neighbours said that the father-daughter duo, originally from Malaysia, had been living in the stall virtually day and night. They had never seen the girl go to school. The neighbours said that the 15-year-old did not leave the stall, even to wash or go to the toilet, and the gate was shut during the day as well as at night. 

Also Read | Vietnam Teacher Sits On 5-Year-Old Boy, Hits Face And Forces Him To Eat Oranges

Mr Tan was protective of his daughter and never let her talk to other stall owners, they said. He also rejected offers to help. “He did not beat or abuse her, but treated her like a pet,” a neighbour said, per SCMP. The neighbours also said that the duo used to have a home, but was not known why they moved to the stall. 

Singapore’s National Environment Agency reported the case to the Ministry of Social and Family Development, following which the girl was sent to hospital for a check-up. Neighbours said they spotted Mr Tan a few times after his daughter was hospitalised. He reportedly told them he would be taking her back to Malaysia. 

The police are investigating the father for alleged child abuse. The probe is yet to reach a conclusion. 

Notably, under Singapore’s Children and Young Persons Act, ill-treatment of a child aged 16 or younger can result in a fine of up to USD 5,900 and a maximum jail term of eight years.
 

The man accused of killing his son by forcing the six-year-old into a bruising treadmill workout has claimed that it was pneumonia, and not the gym session, that killed the boy. According to New York Post, Christopher Gregor’s lawyer said during a court hearing that treadmill running “has nothing to do” with the death of Corey Micciolo. The incident took place in March 2021 and Gregor, 31, was arrested three months later. He is being held in the Ocean City Jail without bond.

The case gained significant attention after a disturbing video was presented in the court in Ocean County that showed Gregor forcing the six-year-old boy to run on a treadmill at high speed and incline at the Atlantic Heights Clubhouse fitness centre, allegedly because he thought his son was “too fat”.

NEW: Mother breaks down in court as she watches her son’s father abuse her child by making him run on the treadmill because he was “too fat.”

New Jersey father Christopher Gregor is accused of killing his 6-year-old son Corey Micciolo.

New footage shows the boy repeatedly face… pic.twitter.com/aVKknkOGd5

— Collin Rugg (@CollinRugg) May 1, 2024

Gregor increased the speed and incline in treadmill settings until the boy flew off. In the surveillance video, he is seen picking the boy up and putting him back on the treadmill, but Corey falls again and continues to struggle until his father lowers the speed and gives the boy a break.

On Thursday, Gregor’s lawyer told the court, “When you see that video, you’re going to be horrified. You’re going to be mortified.”

“(But) Corey’s death had absolutely nothing to do with that treadmill,” the Post further quoted the lawyer as saying.

The prosecutors, however, disagreed with the argument and said it was his father’s chronic abuse that left Corey with internal injuries that ultimately cost him his life.

The boy’s mother Breanna Micciolo also took to the witness stand and said she noticed bruises on her son. “He had a bruise on his forehead. It was very odd-shaped. It looked like an impression. He had one on his chest. It looked like a scrape,” she told the jury.

Ms Micciolo was already in a custody battle with Gregor because she feared for her son’s life.

Murder charge was formally slapped on Gregor in March 2022 and the trial began.

Former YouTube vlogger Ruby Franke, known for offering parenting advice to millions online, received a sentence of up to 60 years in prison on Tuesday after pleading guilty to child abuse charges, according to The Independent.

Franke, a mother of six, broke down in tears while addressing the court before Judge Richard Christofferson sentenced her to four consecutive terms of one to 15 years. These sentences stem from four counts of child abuse to which she pleaded guilty in December, according to the newspaper.

The charges relate to the abuse of two of Franke’s children, aged nine and 11 at the time, allegedly involving psychological manipulation, food deprivation, and isolation. Prosecutors described the victims’ living conditions as “concentration camp-like” and called her a significant threat to the community.

Also Read | Parenting Advice YouTuber Ruby Franke Pleads Guilty To Child Abuse

Franke’s former business partner, Jodi Hildebrandt, who was also involved in the abuse, received the same sentence.

“The children were regularly denied food, water, beds to sleep in, and virtually all forms of entertainment,” Utah prosecutor Eric Clarke said.

In court, Franke was in tears following the sentence. She apologised to her children and said: “I was so disoriented that I believed dark was light and right was wrong.”

“I was led to believe that this world was an evil place, filled with cops who control, hospitals that injure, government agencies that brainwash, church leaders who lie and lust, husbands who refuse to protect, and children who need abuse,” she said.

In August 2023, Ruby Franke and Jodi Hildebrandt were arrested following an incident involving Franke’s undernourished 12-year-old son. The child escaped through a window at Hildebrandt’s residence in Ivins, Utah, seeking assistance from a neighbor by requesting food and water. Police records indicate that the child sustained lacerations from being restrained with a rope.