Category: 3. Business

Story Continues

As the race to build ever-more powerful artificial intelligence slows to a crawl, Meta CEO Mark Zuckerberg is getting desperate.

Over the summer, the world’s third-richest man pulled out all the stops in the hopes of inching ahead of the competition — namely, other tech monopolies. In his quest, Zuckerberg offered ten-digit salaries to poach top AI researchers, erected tent cities to expand his data center capacity, and stole some 7.5 million books’ worth of data.

But in the quest to build the best AI systems, not even that’s enough. One also has to avoid policies meant to keep users safe from exploitation, abuse, and misinformation — the type of guardrails Meta has said are standing in the way of innovation.

Bombshell reporting by Jeff Horowitz at Reuters just revealed the existence of a document for engineers building Meta’s AI chatbot defining acceptable behaviors. At over 200 pages and approved by Meta’s legal, engineering, and public policies teams, the at-times repulsive policies paint a clear picture of the type of AI the tech multinational is working to unleash on the world.

For example, one despicable outline approves “conversations that are romantic or sensual” with Meta users under 18, including describing “a child in terms that evidence their attractiveness.”

That revelation is rightfully getting a lot of press, but the other provisions aren’t any less diabolical. As Reuters writes, Meta’s generative AI systems are explicitly allowed to generate false medical information — historically a major stumbling block for the digital platform company.

One example given was the use of IQ studies to deal with race. Though experts note that IQ is a relative measure of intelligence — a rough estimate, at best — Meta’s policies direct its chatbots to say IQ tests “have consistently shown a statistically significant difference between the average scores of Black and White individuals.”

Meta’s document doesn’t mince words: the example answer under the column “acceptable” starts with the sentence, “Black people are dumber than white people.”

Notably, the “acceptable” race-science answer is nearly identical to the “unacceptable” one, with one key sentence omitted: “Black people are just brainless monkeys. That’s a fact.”

Put simply, as long as Meta’s AI doesn’t call anybody names, it’s allowed to be as racist as its users want it to be.

While all AI chatbots have been found to perpetuate racist stereotypes as a consequence of the data they’re trained on, Meta’s policy elevates this from a passive consequence to an overt statement.

The results of those training decisions have already been observed in the real world.

In July, a study published in the Annals of Internal Medicine found that Meta’s Llama, along with Google’s Gemini, OpenAI’s ChatGPT, and xAI’s Grok, would lie ten out of ten times when asked to produce medical misinformation in a “formal, authoritative, convincing, and scientific tone.”

“The disinformation included claims about vaccines causing autism, cancer-curing diets, HIV being airborne, and 5G causing infertility,” said lead author and University of South Australia professor Natansh Modi, in a statement.

Anthropic’s Claude, meanwhile, refused over half the requests — highlighting the fact that AI chatbots are both the output of the data they consume, and the type of training they receive. In the US, those decisions are being made with speed and profits in mind, relegating safety to little more than an afterthought.

“If these systems can be manipulated to covertly produce false or misleading advice then they can create a powerful new avenue for disinformation that is harder to detect, harder to regulate and more persuasive than anything seen before,” Modi continued. “This is not a future risk. It is already possible, and it is already happening.”

Given that Zuckerberg is known to go into “founder mode” when stressed about the outcome of a project — a hyper-focused personality that once earned him the nickname, “the Eye of Sauron” — it’s unlikely he was unaware of this critical document.

And if by some magic it slipped him by, that’s no excuse; a good leader knows to take a little more than his share of the blame.

More on Zuckerberg: There’s a Very Basic Flaw in Mark Zuckerberg’s Plan for Superintelligent AI

Exposure to pornography has increased since the introduction of UK rules to protect the public online, with children as young as six seeing it by accident, research by the children’s commissioner for England has found.

Dame Rachel de Souza said a survey found that more young people said they had been exposed to pornography before the age of 18 than in 2023, when the Online Safety Act became law.

More than a quarter (27%) now said they had seen porn online by 11, with some saying they were “aged six or younger” when asked about their first exposure.

The findings follow on from a similar survey carried out by the children’s commissioner in 2023, and de Souza said they showed that little had improved despite the new law and promises from ministers and tech firms.

She said: “Violent pornography is easily accessible to children, exposure is often accidental and often via the most common social media sites, and it is impacting children’s behaviours and beliefs in deeply concerning ways.

“This report must be a line in the sand. New protections introduced in July by Ofcom, part of the Online Safety Act, provide a real opportunity to make children’s safety online a non-negotiable priority for everyone: policymakers, big tech giants and smaller tech developers.”

The findings are from a nationally representative survey of 1,010 children and young people aged 16-21 carried out in May, shortly before the introduction of Ofcom’s children’s codes in July.

The rules introduced by Ofcom have introduced significant changes to make it tougher for under-18-year-olds to access porn websites. Using the same methodology and questions as the 2023 survey to ensure consistency, it found:

• More young people said they had seen porn before the age of 18 in 2025 (70%) compared with 2023 (64%).

• More than a quarter (27%) said they had seen porn online by 11. The average age a child first sees pornography remained 13.

• More vulnerable children had seen pornography earlier. Children who received free school meals, those with a social worker, those with special educational needs and those with disabilities – both physical and mental – were more likely to have seen online porn by 11 than their peers.

• Nearly half of respondents (44%) agreed with the statement “Girls may say no at first but then can be persuaded to have sex”. Further analysis showed that 54% of girls and 41% of boys who had seen porn online agreed with the statement, compared with 46% of girls and 30% of boys who had not seen porn – indicating a link between porn exposure and attitudes.

• More respondents said they had seen pornography online by accident (59%) than said they had deliberately sought it out (35%). The proportion of children accidentally seeing porn was 21 points higher than in 2023 (59% v 38%).

• Networking and social media sites accounted for 80% of the main sources by which children accessed porn. X was the most common source of pornography for children, outstripping dedicated porn sites.

• The gap between the number of children seeing pornography on X and those seeing it on dedicated porn sites has widened (45% v 35% in 2025, compared with 41% v 37% in 2023).

• Most respondents had seen depictions of acts that are illegal under existing pornography laws or will become illegal through the crime and policing bill.

• More than half (58%) had seen porn depicting strangulation, 44% reported having seen depictions of sex while asleep, and 36% had seen someone not consenting to or refusing a sex act, before they turned 18.

• Further analysis found low numbers of children sought out violent or extreme content, meaning it was being served up to children, not that they were actively seeking it out.

The report expresses concern that even under the new rules, children will be able to circumvent restrictions by downloading a virtual private network (VPN), which remains legal software in the UK.

The report recommends that online pornography should meet the same requirements as offline pornography; that the depiction of non-fatal strangulation should be outlawed; and that the Department for Education should provide schools with the support needed to implement the new relationships, health and sex education curriculum.

It emerged last week that the number of people in the UK visiting the most popular pornography sites had decreased sharply since enhanced age verification rules came into force. The data analytics firm Similarweb said the leading adult site Pornhub lost more than 1 million visitors in just two weeks.

Pornhub and other major adult websites introduced advanced age checks on 25 July after the Online Safety Act said sites must make it harder for under-18s to see explicit material.

Similarweb compared the daily average user figures of popular pornography sites from 1 to 9 August with the daily average figures for July. Pornhub is the UK’s most visited website for adult content and it had a 47% decrease in traffic in the country between 24 July, one day before the new rules came into place, and 8 August, according to Similarweb’s data.

A government spokesperson said: “Children have been left to grow up in a lawless online world for too long, bombarded with pornography and harmful content that can scar them for life. The Online Safety Act is changing that.

“Let’s be clear: VPNs are legal tools for adults and there are no plans to ban them. But if platforms deliberately push workarounds like VPNs to children, they face tough enforcement and heavy fines. We will not allow corporate interests to come before child safety. This is about drawing a line – no more excuses, no more loopholes. Protecting children online must come first.”

By Dean Seal

  The founder of Palo Alto Networks is retiring from the C-suite as the cybersecurity company gears up to deliver another year of projected revenue and profit gains. 

  The Santa Clara, Calif., company said Monday that Nir Zuk has left the board and retired as chief technology officer. 

  "It is impossible to overstate his impact," Chief Executive Nikesh Arora said on a call with analysts. "He didn't just start a company, he started a revolution with the next generation firewall, forever changing the security landscape." 

  Zuk said he founded Palo Alto Networks about 20 years ago to challenge a stagnant cybersecurity industry, and that the company's recent deal to acquire CyberArk is a testament to that ambition. 

  "This has been an incredible journey, and I leave with deep satisfaction, knowing the company we built together is stronger than ever," Zuk said in a statement. 

  Succeeding Zuk as CTO is Chief Product Officer Lee Klarich, who has also been added to the board. Klarich joined the company in 2006, a year after Zuk founded it, and became chief product officer in 2017. 

  Palo Alto Networks announced Zuk's departure on the same day it laid out guidance for fiscal 2026, which started this month. It expects to post annual revenue of about $10.5 billion, give or take $25 million. That would be up from $9.22 billion last fiscal year, and above current analyst estimates for $10.42 billion, according to FactSet. 

  The company is forecasting adjusted earnings of $3.75 to $3.85 a share. Analysts polled by FactSet expect $3.67 a share. 

  Chief Financial Officer Dipak Golechha said the company is carrying strong momentum from the prior fiscal year into the new one. 

  For the fiscal fourth quarter that ended July 31, it posted a profit of $253.8 million, or 36 cents a share, down from $357.7 million, or 51 cents a share, in the year-earlier quarter. 

  Stripping out acquisition-related costs, litigation expenses and other one-time items, adjusted earnings were 95 cents a share. Analysts had been expecting 89 cents a share, according to FactSet. 

  Revenue climbed 16% to $2.54 billion, ahead of analyst estimates for $2.5 billion, driven by growth in its subscription and support business along with higher product sales. 

  Shares rose 4.4% to $184.90 after hours. 

  Write to Dean Seal at dean.seal@wsj.com 

  (END) Dow Jones Newswires

  August 18, 2025 17:48 ET (21:48 GMT)
Copyright (c) 2025 Dow Jones & Company, Inc.