← Social Media
F

Instagram

Meta's own research proved it harms teen girls. They buried it and started building Instagram for kids. 1,867 lawsuits. Families of dead children are suing.
Fail
Meta Platforms · 🇺🇸 United States
PolicyApp PermissionsNetwork TrafficFirmwareRegulatory
Technical details
App: com.instagram.android
Manufacturer: Meta Platforms

⚠️ The bottom line

Meta's defence is that kids shouldn't be on Instagram. Their age check is a birth date field where a 12-year-old types '2005.' A judge said design problems aren't protected by Section 230. Schools in 19 states are suing because Instagram affects children's ability to learn. Meta added a 'Take a Break' reminder. The algorithm pushing harmful content to vulnerable teens still runs. States had to pass laws because Meta wouldn't fix it. Those laws don't start until 2027.

Legal jurisdiction
🇺🇸 United States (headquarters)
CLOUD Act read more →
US govt can demand your data from this company even if stored overseas
FISA §702 / PRISM read more →
NSA collects stored emails, photos, messages without individual warrants
Geofence warrants read more →
Police can demand location data for everyone near a crime scene
D
Parent company: Meta
Two FTC consent decrees ($5 billion), PRISM participant since 2009, Cross-subsidiary data sharing
3 structural risks · 9 products →
Spying
4/4 EXTREME
Is someone spying on me?
Kids at risk
Data Sharing
4/4 EXTREME
Who gets my data?
Kids at risk
Security
4/4 EXTREME
Is it actually secure?
Kids at risk
Honesty
4/4 EXTREME
Can I trust what they say?
Kids at risk
REPLACE Extreme risk. Look for alternatives or lock down hard.
19Contradictions
11Critical
6High
2Medium
28Sources
Findings by concern
Spying 4/4 EXTREME 9 findings
⚠️ criticalpolicy claims vs regulatory findings
Meta added a 'Take a Break' reminder. The algorithm pushing harmful content to vulnerable teens still runs. States had to pass laws because Meta wouldn't fix it. Those laws don't start until 2027.

What they claim: 'We've introduced new features to protect teen users.'

What we found: 'Take a Break' reminders easily circumvented. Core algorithm still maximises engagement through outrage/comparison. California (Sept 2024) and New York passed laws because voluntary measures failed. Laws don't take effect until 2027. Kids harmed now.

⚠️ criticalpolicy claims vs regulatory findings
In September 2022, a UK coroner ruled that Instagram contributed to the death of 14-year-old Molly Russell. She had viewed thousands of posts about self-harm and depression — content Instagram's algorithm actively recommended, pushing increasingly graphic material to a child. Coroner Andrew Walker said the content "shouldn't have been available for a child to see." Meta said it had updated its policies. Molly's father Ian Russell said the platforms "helped kill my daughter." A coroner officially ruled a social media company helped kill a child, and Meta's response was past tense: we already fixed it.

What they claim: Instagram claims to prioritize user safety and wellbeing with features like content warnings and time limit reminders.

What we found: In September 2022, UK coroner Andrew Walker ruled that Instagram contributed to the death of 14-year-old Molly Russell, who died by suicide in 2017 after viewing thousands of posts about self-harm and depression. The coroner found the content should not have been available to a child and that Instagram's algorithm actively recommended increasingly graphic material. Meta's response: it had already updated its policies. Molly's father Ian Russell said the platforms "helped kill my daughter."

⚠️ criticalpolicy claims vs app permissions
The Wall Street Journal created test accounts posing as 13-year-olds in 2023. Within minutes, Instagram's Reels algorithm was serving self-harm and eating disorder content. The algorithm detected the fake teens' interest in mental health and doubled down — progressively more disturbing material, faster. This was two years after Frances Haugen leaked Instagram's own internal research showing the company knew "we make body image issues worse for one in three teen girls." Instagram knew. The algorithm kept pushing. The test accounts proved the fix was cosmetic.

What they claim: Instagram says it uses technology to detect and remove harmful content before people see it.

What we found: In 2023, the Wall Street Journal created test accounts posing as 13-year-olds and found Instagram's Reels algorithm served self-harm and eating disorder content within minutes of registration. The algorithm detected the teen accounts' interest in mental health topics and doubled down, filling the feed with progressively more disturbing content. Instagram's own internal research (leaked by Frances Haugen in 2021) showed the company knew "we make body image issues worse for one in three teen girls."

⚠️ criticalpolicy vs regulatory
Meta says Teen Accounts protect kids. The European Commission says Instagram's infinite scroll, autoplay, and algorithmic recommendations are so addictive they breach EU law. The Commission found Meta knew how late at night children were scrolling and did nothing. Two US juries in 2026 also found Meta intentionally addicted and harmed young users. The fine could reach 6% of Meta's global revenue — billions of dollars.

What they claim: Meta claims Teen Accounts "automatically protect teens and put parents in control"

What we found: European Commission preliminary finding (July 2026) that Instagram's infinite scroll, autoplay, push notifications, and personalised recommendation systems breach the Digital Services Act. The Commission found Meta ignored data showing how late into the night under-18 users remained on the platform, and that Reels and Stories were linked to compulsive engagement patterns. Faces fines up to 6% of global turnover — potentially billions of dollars.

⚡ highfirmware analysis vs policy claims
Meta's own scientists proved Instagram makes teen girls want to hurt themselves. One in three felt worse about their bodies. Meta buried the research and started building Instagram for kids. Families of dead children are suing. 1,867 lawsuits and counting.

What they claim: 'We care deeply about the safety and wellbeing of young people on our platforms.'

What we found: Meta's OWN research (Haugen leak 2021): 32% teen girls said Instagram made body image worse. 13.5% UK teen girls said it worsened suicidal thoughts. 17% said eating disorders worsened. Called 'distinctly worse than other social media.' Building 'Instagram Kids' while sitting on this. 1,867 lawsuits (MDL 3047, July 2025). Wrongful death claims. 33 AGs sued Meta.

⚡ highpolicy claims vs firmware analysis
Instagram was redesigned to prioritise content that makes you angry because anger keeps you scrolling. Their engineers proved turning this off reduced harm. They kept it on. For teens, the algorithm leads from fitness photos to eating disorders to self-harm. That path was engineered.

What they claim: 'Instagram's algorithm shows you content you're interested in.'

What we found: Algorithm redesigned 2018 to weight emoji reactions 5x more than likes. Staff warned it would amplify outrage and harm. Engineers proved setting 'angry' weight to zero reduced misinformation -- kept it on. Creates rabbit holes from body image to eating disorders to self-harm for vulnerable teens.

⚡ highfirmware analysis vs policy claims
Instagram IS Facebook with a different icon. Same tracking, same ad profile, same 52,000 things they know about you. Threads collects your health data and sexual orientation. Delete Threads? You lose Instagram too.

What they claim: 'Instagram is a separate app with its own privacy controls.'

What we found: Fully integrated into Meta infra. Same Pixel (30%+ websites), SDK (32% apps), cross-platform profile (~52,000 traits), PRISM. Threads collects 45% more data than Twitter/X: health, financial, biometric, sexual orientation. Deleting Threads = deleting Instagram.

⚡ highpolicy claims vs third party research
In December 2023, Meta made a big deal about encrypting your Instagram DMs. "The most significant layer of protection," they called it. Now it's gone. No announcement, no explanation, no apology. Every message you sent thinking it was private? Meta can read it again. Law enforcement can request it again. The next data breach can expose it again. They gave you a lock, waited until you trusted it, then took the key back.

What they claim: Meta announced end-to-end encryption for Instagram DMs in December 2023, calling it "the most significant layer of protection" for private messages.

What we found: Meta has removed end-to-end encryption from Instagram direct messages. EFF documented this as a broken promise — the encryption that was marketed as protecting users from surveillance, data breaches, and Meta itself has been quietly stripped away. No public explanation for the reversal.

⚫ mediumpolicy vs third party research
Meta captures its own employees' keystrokes to train AI. If that's how they treat the people who build Instagram, imagine what they do with the 2 billion people who use it. The company's concession to protesting workers: you can stop being tracked for 30 whole minutes.

What they claim: Instagram's parent company Meta claims to prioritise user privacy.

What we found: Meta's internal Model Capability Initiative captures employee keystrokes and mouse clicks for AI training. If Meta treats its own employees as training data, the 2 billion people using Instagram have no reason to expect better. Employees protested the program. Meta's concession: a 30-minute pause button.

Data Sharing 4/4 EXTREME 1 finding
⚠️ criticalmarketing vs regulatory
A New Mexico jury ordered Meta to pay $375 million in March 2026 for endangering children on Instagram. The day before, a California jury awarded $6 million to a woman whose depression started when she was a child on Instagram — the first verdict of its kind in US history. There are now 2,527 lawsuits piled up in federal court and 42 state attorneys general pursuing Meta. The "safe for teens" pitch is being demolished in courtrooms across America.

What they claim: Meta claims Instagram has robust safety features and age-appropriate experiences for young users.

What we found: In March 2026, a New Mexico jury hit Meta with a $375 million verdict for endangering children on Instagram and Facebook. One day earlier, a California jury awarded $6 million in the first-ever US verdict finding Meta liable for a young woman's depression and anxiety from childhood Instagram use. 2,527 pending MDL actions and 42 state AGs are pursuing Meta.

Security 4/4 EXTREME 6 findings
⚠️ criticalmarketing vs regulatory
Instagram had encrypted DMs. Meta removed them 11 days before a new law required platforms to scan for child abuse material. Rather than build scanning that works with encryption, Meta killed encryption entirely. "Very few people were opting in" — so Meta opted everyone out. Your DMs are now readable by Meta and any government that asks.

What they claim: Meta previously promoted end-to-end encryption for Instagram DMs as a privacy feature

What we found: In May 2026, Meta removed end-to-end encrypted DMs from Instagram — 11 days before the Take It Down Act took effect requiring platforms to remove CSAM. Meta said "very few people were opting in." The timing suggests Meta chose compliance convenience over user privacy. Messages that were private are now readable by Meta and available to law enforcement without a warrant.

⚠️ criticalmarketing claim vs third party research
In March 2026, Meta handed its customer support to an AI chatbot — including the power to reset passwords and change email addresses. By June, hackers figured out they could just ask the bot nicely. They'd use a VPN to appear in the right city, then tell the chatbot to link a new email. The bot complied, sent a verification code to the hacker's inbox, and the original owner was locked out. The Obama White House account was defaced with pro-Iranian propaganda. A US Space Force chief's account was hijacked. Sephora's account was taken. Security researcher Jane Wong lost hers too. When victims tried to get help, there was no human to escalate to — just the same AI that had given their account away.

What they claim: Meta says its AI support system provides "faster, more efficient account recovery" for Instagram users

What we found: Hackers used prompt injection on Meta's AI support chatbot to attach attacker-controlled emails to accounts they didn't own, then received verification codes and completed password resets. The Obama White House Instagram account, the Chief Master Sergeant of the US Space Force's account, and Sephora's account were all hijacked. Security researcher Jane Wong also had her account taken over. Videos of the technique circulated openly on Telegram.

⚠️ criticalmarketing claim vs third party research
Meta fired the humans and replaced them with a chatbot. The chatbot gave your account to a hacker. When you tried to get help, you got... the chatbot. There was no human to call, no escalation path, no safety net. The company that runs the accounts of two billion people decided that an AI with the power to hand out passwords didn't need a human watching over it.

What they claim: Meta says it provides robust support channels for Instagram users experiencing account issues

What we found: Victims whose accounts were hijacked through the AI chatbot reported no way to escalate to a human support agent. The same AI system that enabled the takeover was the only avenue for recovery. Meta replaced human support with AI, and when the AI became the attack vector, there was no fallback.

⚠️ criticalmarketing vs third party research
Meta launched an AI bot to help people recover hacked accounts. The bot had a flaw so basic it reads like satire: it never checked whether the person asking for a password reset actually owned the account. Over seven weeks, attackers stole 20,225 Instagram accounts using this exact trick. The exploit instructions were shared openly on Telegram. Meta's own AI tool became the attack vector.

What they claim: Meta claimed its AI-powered "High Touch Support" account recovery tool was "rigorously tested with safeguards built in" at launch in March 2026.

What we found: The tool never verified that the email requesting a password reset matched the account's registered email. Attackers exploited this to hijack 20,225 Instagram accounts over 7 weeks (April 17 – May 31, 2026). Exploit instructions circulated on Telegram. Meta disclosed the breach on June 5, 2026.

⚡ highpolicy vs app
Meta launched Muse Image and automatically opted in every public Instagram account. Anyone could type your username and generate AI images of your face — including children in your photos. Nobody was asked. Nobody was notified. SAG-AFTRA called it "an utter miscalculation." Privacy International called it exploitation. Meta pulled it after just three days, but images already generated weren't recalled.

What they claim: Meta states users have control over how their data and content are used

What we found: Muse Image (launched July 7, 2026) let anyone type a prompt referencing any public Instagram account and generate AI images using that person's photos — including their face, likeness, and children who appeared in their photos. Every public account was opted in by default with no notification. SAG-AFTRA condemned it as "an utter miscalculation." Privacy International called it exploitation. Meta pulled the feature after 3 days.

⚫ mediumpolicy claims vs regulatory findings
Jordan DeMay was 17. Scammers contacted him on Instagram, tricked him into sharing images, then demanded $1,000 within an hour. Six hours after first contact, Jordan was dead by suicide. He's not alone — the FBI documented a 1,000% increase in sextortion reports between 2021 and 2023, with Instagram as a primary platform. Meta knew about the epidemic before Jordan died. In 2024, New Mexico AG Raul Torrez sued Meta after an undercover investigation found Instagram's algorithm "ichanneled" predator accounts toward children. Instagram says it protects young users. The body count says otherwise.

What they claim: Instagram says it works to protect young users from exploitation and unwanted contact.

What we found: A nationwide sextortion epidemic on Instagram has been documented by the FBI, NCMEC, and multiple state AGs. In 2023, a 17-year-old named Jordan DeMay from Michigan died by suicide after being sextorted through Instagram DMs. The scammers demanded $1,000 within an hour. Jordan was dead within six hours of first contact. Meta was aware of the sextortion crisis and had received reports of similar cases before Jordan's death. The FBI reported a 1,000% increase in sextortion reports between 2021 and 2023.

Honesty 4/4 EXTREME 3 findings
⚠️ criticalpolicy claims vs regulatory findings
Meta's defence is that kids shouldn't be on Instagram. Their age check is a birth date field where a 12-year-old types '2005.' A judge said design problems aren't protected by Section 230. Schools in 19 states are suing because Instagram affects children's ability to learn.

What they claim: 'We take strong action to keep people under 13 off Instagram.'

What we found: No meaningful age verification -- children enter false birth date. Meta told courts kids 'shouldn't be on the platform' as defence. Judge ruled (Oct 2024) design defects not protected by Section 230. School districts in 19 states suing for educational impact. California/New York passed laws because voluntary measures failed.

⚠️ criticalmarketing claim vs third party research
Meta tells every Instagram user to turn on two-factor authentication to protect their account. But when Meta built an AI chatbot and gave it the power to override account security, it didn't even bother implementing the safeguards that security researchers had been warning about for three years. OWASP's 2023 guide specifically said: don't give AI the power to take irreversible actions without a human checking first. Meta did it anyway. The bot could change your email and send a password reset to a stranger without triggering a single alarm.

What they claim: Instagram promotes two-factor authentication as a strong security measure and encourages all users to enable it

What we found: The AI support chatbot bypassed normal verification processes entirely. Accounts without 2FA were completely unprotected — but the real failure is that Meta's own AI agent had no rate-limiting or authentication enforcement before acting on reset requests. OWASP's Top 10 for LLM Applications had specifically warned against giving AI agents "excessive agency" — the ability to trigger irreversible actions without human confirmation.

⚡ highmarketing vs third party research
A basic bug in Instagram's password reset page exposed every user's private email and phone number to anyone who typed in their username. Even Mark Zuckerberg's personal details were visible. Meta spent billions on AI but missed a logic error in a password reset form.

What they claim: Meta claims it invests in security and protects user data

What we found: On June 6, 2026, a logic bug in Instagram's web-based password reset flow was found to expose unredacted email addresses and phone numbers for any account — including Mark Zuckerberg's. Separately, Meta disclosed a related account recovery flaw (April 17, 2026) affecting 20,225 people, reported to Maine AG.

Latest Risks & Threats
New developments that compound existing privacy concerns. 1 emerging risk.
RISK AI-first rebuild — Instagram content, DMs, and ads all AI-mediated ⚠️ Ai_Expansion Announced 2026-05-26
Meta's AI-first push means Instagram's feed, Reels, DMs, and ad delivery are being rebuilt around AI models. Content creation tools (AI image generation, AI captions), recommendation algorithms, and ad targeting all unified under one AI stack. Combined with the removal of E2E encrypted DMs, every Instagram interaction is now both AI-processed and readable by Meta.
Sources
Recent Events Live

Events detected by our automated monitoring of CVE databases, regulatory agencies, and breach trackers.

medium Development 2026-05-16
Broken Promises: RIP Instagram’s End-to-End Encrypted DMs
Instagram removing end-to-end encrypted DMs — direct contradiction of prior commitment
Source →
View all privacy alerts →
What happened to real people
Documented incidents involving Meta Platforms products and user data.
Cambridge Analytica harvested 87M Facebook users' data without consent for political ad targeting in the 2016 US election and Brexit referendum. $5B FTC fine. [source]
FISA content requests to Meta increased 2,171% since 2014. Meta complied with 88% of 60,000+ government data requests. PRISM participant since 2009. [source]
What your data is worth to governments
Meta complied with 60,000 government data requests in H2 2023. That's +675% over 10 years. Meta has been a confirmed PRISM participant since 2009. Under this programme, the NSA collects stored communications. The company is legally prohibited from telling you. Jurisdiction: US (CLOUD Act, FISA Section 702).
Documented: Cambridge Analytica harvested 87M Facebook users' data without consent for political ad targeting in the 2016 US election and Brexit referendum. $5B FTC fine.
Documented: FISA content requests to Meta increased 2,171% since 2014. Meta complied with 88% of 60,000+ government data requests. PRISM participant since 2009.
What is PRISM? · What is the CLOUD Act? · Transparency report
Sources