Meta's Policy Shift Sparks Surge in Harmful Content Across Facebook
Meta's latest Integrity Report reveals an unsettling trend: a notable uptick in violent content and online harassment on Facebook. The rise comes in the wake of a strategic pivot by Meta to ease moderation policies, aimed at reducing enforcement errors and encouraging greater political expression.
This is the company's first major report since it implemented these changes in January 2025. It provides insight into how the softer approach is playing out across its major platforms—Facebook, Instagram, and Threads. While Meta's intention was to make moderation more balanced and less prone to error, the results are raising red flags.
Harmful Content on the Rise
The numbers are telling. Meta reported that the volume of violent and graphic content on Facebook grew from 0.06–0.07 percent in late 2024 to 0.09 percent in the first quarter of 2025. While the percentages may appear small, the actual volume is substantial given Facebook's massive user base.
The company also noted an increase in bullying and harassment content, largely attributed to a surge in March. "There was a small increase in the prevalence of bullying and harassment content from 0.06–0.07 percent to 0.07–0.08 percent on Facebook due to a spike in sharing of violating content in March," the report stated.
These reversals are particularly concerning as they follow years of gradual decline in harmful content, casting doubt on the effectiveness of Meta's updated enforcement strategy.
Less Content Being Removed
Alongside the rise in harmful content, there's been a significant drop in content removals. In Q1 2025, Meta took action on just 3.4 million pieces under its hate speech policy—the lowest since 2018. Spam removals halved, plunging from 730 million at the end of 2024 to 366 million in early 2025. Even fake account removals dipped from 1.4 billion to 1 billion.
Meta's revised approach focuses only on the most egregious content, such as child exploitation and terrorism, leaving more nuanced or controversial posts untouched. Topics like immigration, gender identity, and race, previously subject to stricter scrutiny, are now categorised under political discourse and granted wider leeway.
Redefining Hate Speech
The company has also narrowed its definition of hate speech. Where it once included contemptuous or exclusionary language, the new policy focuses solely on direct attacks and dehumanizing language. As a result, content previously flagged for expressing inferiority or exclusion now slips through the cracks.
This change is part of Meta's effort to minimise over-enforcement, but has raised concern among experts who warn that harmful rhetoric may go unchecked under the new rules.
A Shift in Fact-Checking
Another significant change came in early 2025 when Meta ended its U.S.-based third-party fact-checking partnerships. In their place, the company launched a user-driven initiative called Community Notes. These are now live across Facebook, Instagram, Threads, and even Reels and Threads replies.
Although Meta has yet to provide data on the effectiveness of Community Notes, some experts have expressed doubts about its reliability. They worry the system, heavily dependent on crowd input, could be vulnerable to bias or manipulation without editorial oversight.
Meta Claims Progress Despite Warnings
Despite these concerning trends, Meta claims its new moderation model is showing success in reducing enforcement mistakes. According to the company, moderation errors in the U.S. dropped by roughly 50 percent from Q4 2024 to Q1 2025. However, Meta has not clarified how this figure is calculated, though it promises to provide clearer metrics in future reports.
The company says it is striving to 'strike the right balance' between being too lenient and too aggressive in enforcement.
Teen Safety Still a Focus
One area where Meta continues to maintain strict moderation is content directed at teenagers. The company is rolling out Teen Accounts across its platforms to better filter content and shield younger users from bullying, violence, and inappropriate material.
While the company remains committed to teen safety, its broader approach to content moderation appears to be under increasing scrutiny. With harmful content trending upward and enforcement actions in decline, Meta may soon face pressure to reconsider how it polices the world's largest social media platforms.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Time of India
4 hours ago
- Time of India
Under fire, Sarvam AI co-founder says worries about Indic GenAI model premature
Chennai/Bengaluru: India's latest home-grown generative AI language model, Sarvam-M, has drawn fire from sections of the developer community for what they describe as "under-whelming" performance. Tired of too many ads? go ad free now But Pratyush Kumar, co-founder of Sarvam AI, insists the scepticism is premature and betrays a misunderstanding of how AI frontier models mature. "The ecosystem is early and people are worried too early," he tells TOI. "We are scrambling amongst ourselves when the world moves fast. We want to create an AI ecosystem where more people can positively collaborate." Released this month, the relatively small 24-billion-parameter Sarvam-M model was trained to reason across ten Indian languages while tackling maths and coding tasks. Kumar says benchmarks on Hugging Face (a platform and open-source library primarily used for leveraging machine learning models) show the model matching or outscoring popular open-source rivals (like Meta's Llama, Mistral Small and Gemma 3) in mathematics, programming and Indic-language comprehension. "With this we want to show that we cracked post-training (process of refining and optimising a machine learning model after its initial training phase) problems and our methodology is comparable with other models," he explains. "We open-sourced this because we want to show that such a model can be built and encourage other people to do it." Much of the social-media push-back has centred on relatively modest early-stage download numbers and the perception that Sarvam-M offers few breakthrough capabilities. Kumar counters that India's sovereign-AI ambitions demand more than one blockbuster release. "These things involve both scientific explorations and resource consumption," he says. Tired of too many ads? go ad free now "I think we are on the path to building state-of-the-art models. " Sarvam AI is the first startup chosen to build a frontier model under the government's IndiaAI Mission, which is funding compute, data and research partnerships to reduce reliance on overseas platforms. Although the latest model is a private effort separate from the IndiaAI Mission, Kumar says the initiative will benefit everyone. He declines to give a timeline for the AI Mission-backed foundation model, noting that the company has yet to receive graphics-processing units (GPUs) from government suppliers. "We will open-source the foundational model," he says, but warns that schedules depend on hardware access and collaborative research cycles. Industry weighs in Seasoned AI practitioners say early criticism overlooks the scale of what Sarvam is attempting. "Building a 24-billion-parameter model in India is not easy, especially when deep research isn't encouraged in most universities or companies," says Jaspreet Bindra, co-founder of consultancy AI&Beyond. "Sarvam-M demonstrates robust multilingual reasoning by supporting ten Indian languages – no other model in the world has such a strong Indic component. " Sourabh Deorah, CEO & co-founder of an AI-powered employee engagement and rewards platform, says that as someone deeply involved in machine learning, he understands how challenging it is to create a 24-bn parameter model that not only handles reasoning tasks like math and programming but also delivers high-quality performance across multiple Indian languages – many of which have long been underserved in the AI space. Piyush Goel, CEO & founder of IT consulting company Beyond Key, says that the new model's potential to drive agentic AI in education, healthcare, and automation is exciting. Agentic AI is a type of AI that makes decisions and takes actions based on context and objectives without constant human intervention. Karthikeyan G, senior director of engineering architecture at software company Ascendion, says Sarvam-M's architecture will enable AI agents to interact among themselves (to take complex decisions) thanks to the standardised protocols being used. This will be crucial for the next stage of the AI wave.
Time of India
7 hours ago
- Time of India
This is the first ever picture clicked on a cell-phone!
Taking a photo today is second nature—you pull out your phone, tap the screen, and capture the moment in an instant. It's casual, seamless, and deeply woven into the fabric of everyday life. Tired of too many ads? go ad free now We don't think twice before sending a selfie, sharing a sunset, or preserving memories with a single click. But what we now take for granted is the result of years of innovation, experimentation, and a dash of tech magic. B efore smartphones and cloud storage, before Instagram and camera rolls, there was one moment—one photograph—that started it all. And it happened on June 11, 1997, in a hospital room in California. How did the first ever picture come to be? On June 11, 1997, engineer and tech entrepreneur Philippe Kahn sat in the maternity ward of Sutter Maternity Center in Santa Cruz, California. His wife was in labor, and as he waited for their daughter to arrive, Kahn decided he didn't just want to take a photo—he wanted to share it instantly. But there were no smartphones then. No Instagram. No instant sharing. So Kahn improvised. Using a Motorola StarTAC flip phone, a Casio QV digital camera that shot low-res 320x240 pixel images, and a Toshiba 430CDT laptop, he built a system from scratch right there in the hospital. The setup was wired so that when he took a photo, it would automatically upload the image to his web server, then send out email alerts to friends and family with a link to view it online. This wasn't just a photo—it was the first time an image was captured and sent directly from a mobile phone. Kahn had already been working on a concept called 'Picture Mail', a vision for sending photos instantly via a server-based system. As he told IEEE Spectrum, he wanted to be the 'Polaroid of the 21st century,' bringing to life a digital version of the instant camera. Tired of too many ads? go ad free now Still, he hadn't developed consumer-ready hardware to make the system easy to use. But time—and necessity—sparked invention. 'I had always wanted to have this all working in time to share my daughter's birth photo,' Kahn said. 'But I wasn't sure I was going to make it.' Luckily for Kahn (and not so luckily for his wife), she was in labor for 18 hours—long enough for him to put his Frankenstein rig together. He had most of the tech on hand, and what he didn't have, an assistant quickly grabbed from a local Radio Shack. As Kahn put it, 'It's always the case that if it weren't for the last minute, nothing would ever get done.' A single photo that sparked a revolution That day, the first photo ever sent from a mobile phone was shared with the world—and though it was a humble 320x240-pixel image of a newborn, it marked the beginning of a technological revolution. We've come a long way since then. Today, more than 1.8 trillion photos are taken each year, mostly from phones that are thinner than a paperback but smarter than computers from the '90s. And it all began with a father, a hospital room, and a dream to share a moment instantly. We've never looked back since.
Scroll.in
9 hours ago
- Scroll.in
Aristotle would scoff at Mark Zuckerberg's suggestion that AI can solve the loneliness epidemic
Mark Zuckerberg recently suggested that AI chatbots could combat social isolation by serving as 'friends' for people experiencing loneliness. He cited statistics that the average American has fewer than three friends but yearns for as many as 15. He was close: According to a 2021 report from the Survey Center on American Life, about half of Americans have fewer than four close friends. Zuckerberg then posited that AI could help bridge this gap by providing constant, personalized interactions. 'I would guess that over time we will find the vocabulary as a society to be able to articulate why that is valuable,' he added. Zuckerberg explaining how Meta is creating personalized AI friends to supplement your real ones: 'The average American has 3 friends, but has demand for 15.' — Roman Helmet Guy (@romanhelmetguy) April 30, 2025 Loneliness and social disconnection are serious problems. But can AI really be a solution? Might relying on AI for emotional support create a false sense of connection and possibly exacerbate feelings of isolation? And while AI can simulate certain aspects of companionship, doesn't it lack the depth, empathy and mutual understanding inherent to human friendship? Researchers have started exploring these questions. But as a moral philosopher, I think it's worth turning to a different source: the ancient Greek philosopher Aristotle. Though it might seem odd to consult someone who lived over 2,000 years ago on questions of modern technology, Aristotle offers enduring insights about friendships – and which ones are particularly valuable. More important than spouses, kids or money In his philosophical text Nicomachean Ethics, Aristotle maintained that true friendship is essential for 'eudaimonia,' a Greek word that is typically translated as 'flourishing' or 'well-being.' For Aristotle, friends are not just nice to have – they're a central component of ethical living and essential for human happiness and fulfillment. 'Without friends, no one would choose to live,' he writes, 'though he had all other goods.' A solitary existence, even one of contemplation and intellectual achievement, is less complete than a life with friends. Friendship contributes to happiness by providing emotional support and solidarity. It is through friendship that individuals can cultivate their virtues, feel a sense of security and share their accomplishments. Empirical evidence seems to support the connection between friendship and eudaimonia. A 2023 Pew Center research report found that 61 per cent of adults in the US say having close friends is essential to living a fulfilling life – a higher proportion than those who cited marriage, children or money. A British study of 6,500 adults found that those who had regular interactions with a wide circle of friends were more likely to have better mental health and be happier. And a meta-analysis of nearly 150 studies found that a lack of close friends can increase the risk of death as much as smoking, drinking or obesity. Different friends for different needs But the benefit of friendship that Aristotle focuses on the most is the role that it plays in the development of virtue. The first tier is what he calls 'friendships of utility,' or a friendship that is based on mutual benefit. Each party is primarily concerned with what they can gain from the other. These might be colleagues at work or neighbours who look after each other's pets when one of them is on vacation. The problem with these friendships is that they are often fleeting and dissolve once one person stops benefiting from the relationship. The second is 'friendships of pleasure,' which are friendships based on shared interests. These friendships can also be transient, depending on how long the shared interests last. Passionate love affairs, people belonging to the same book club and fishing buddies all fall into this category. This type of friendship is important, since you tend to enjoy your passions more when you can share them with another person. But this is still not the highest form of friendship. According to Aristotle, the third and strongest form of friendship is a 'virtuous friendship.' This is based on mutual respect for each other's virtues and character. Two people with this form of friendship value each other for who they truly are and share a deep commitment to the well-being and moral development of one another. These friendships are stable and enduring. In a virtuous friendship, each individual helps the other become better versions of themselves through encouragement, moral guidance and support. As Aristotle writes: 'Perfect friendship is the friendship of men who are good and alike in virtue. … Now those who wish well to their friends for their sake are most truly friends; for they do this by reason of their own nature and not incidentally; therefore their friendship lasts as long as they are good – and goodness is an enduring thing.' In other words, friendships rooted in virtue not only bring happiness and fulfilment but also facilitate personal growth and moral development. And it happens naturally within the context of the relationship. According to Aristotle, a virtuous friend provides a mirror in which one can reflect upon their own actions, thoughts and decisions. When one friend demonstrates honesty, generosity or compassion, the other can learn from these actions and be inspired to cultivate these virtues in themselves. No nourishment for the soul So, what does this mean for AI friends? By Aristotle's standards, AI chatbots – however sophisticated – cannot be true friends. They may be able to provide information that helps you at work, or engage in lighthearted conversation about your various interests. But they fundamentally lack qualities that define a virtuous friendship. AI is incapable of mutual concern or genuine reciprocity. While it can be programmed to simulate empathy or encouragement, it does not truly care about the individual – nor does it ask anything of its human users. Moreover, AI cannot engage in the shared pursuit of the good life. Aristotle's notion of friendship involves a shared journey on the path to eudaimonia, during which each person helps another live wisely and well. This requires the kind of moral development that only human beings, who face real ethical challenges and make real decisions, can undergo. I think it is best to think of AI as a tool. Just like having a good shovel or rake can improve your quality of life, having the rake and the shovel do not mean you no longer need any friends – nor do they replace the friends whose shovels and rakes you used to borrow. While AI may offer companionship in a limited and functional sense, it cannot meet the Aristotelian criteria for virtuous friendship. It may fill a temporary social void, but it cannot nourish the soul. If anything, the rise of AI companions should serve as a reminder of the urgent need to foster real friendships in an increasingly disconnected world.
