Automated trust-and-safety at scale: classifiers, deepfake detection and policy enforcement.
266 articles
French prosecutors summoned Elon Musk to appear in Paris on the 20th for questioning on X platform's issues. The investigation targets the spread of underage pornographic images and rampant deepfake content.
cnBeta (Full RSS) · 152d ago
Apple and Google continue offering mobile apps that generate nonconsensual sexualized images despite explicit policies against them. A Tech Transparency Project report highlights this ongoing issue.
Bloomberg Technology · 157d ago
Roblox will roll out three mandatory age-gated account tiers starting mid-May: Kids (5-8), Select (9-15), and standard (16+), aligned with content maturity labels. Unverified users default to restricted Kids experience without chat or higher-rated games.
Engadget · 159d ago
Instagram is rolling out its 13+ content rating and Limited Content setting internationally, defaulting teens under 18 to age-appropriate feeds without opt-out unless parent-approved. Policies now align with movie ratings, hiding strong language, risky stunts, and harmful behavior content.
Meta Newsroom · 163d ago
UK under-18s reported 394 sextortion cases last year via Report Remove, up 34% from 2024. Campaigners demand nudity-detection tech on phones to combat blackmail after sharing sexual images.
The Guardian Technology · 165d ago
Hongguo Short Drama platform announced governance actions against AI material misuse in short dramas. In Q1, it removed 1,718 non-compliant works, with 670 specifically addressed in a recent review of 15,000 works.
36氪 · 166d ago
ByteDance's short drama platform Hongguo responded to allegations that the AI-generated series Peach Blossom Hairpin used unauthorized facial likenesses. After a 72-hour review following a complaint, the platform found the producer failed to provide evidence of compliant material use.
TechNode · 169d ago
Pangram Labs CEO Max Spero joins the Odd Lots podcast to discuss their software for detecting AI-generated content. The conversation covers implications of AI writing for the internet's future.
Bloomberg Technology · 170d ago
Meta will substantially reduce referencing PG-13 ratings for Instagram Teen Accounts starting April 15, following MPA's cease-and-desist over misleading comparisons. The company updated its blog with a disclaimer highlighting differences between social media and movies.
Engadget · 172d ago
Sony quietly removed 1000+ low-quality AI-generated games from PS Store, mostly from Nostra Games (700 in 2 years via reskins/multi-versions). Targeted trophy hunters; similar spam plagues eShop/Steam.
虎嗅 · 174d ago
YouTube CEO discusses 'AI slop' and distinguishing it from genuine content. The comments suggest vague assurances from the platform.
Digital Trends · 175d ago
Meta's longtime content policy chief Monica Bickert is leaving to teach at Harvard Law School. She oversaw Facebook content policies, user safety, and responded to controversies on political content and teen mental health.
36氪 · 175d ago
OpenAI has abandoned its erotic mode feature for ChatGPT. This is the latest side project the AI startup has ditched in the past week amid several cancellations.
TechCrunch AI · 177d ago
OpenAI reportedly won't add a 'sexy mode' to ChatGPT anytime soon due to investor concerns. The feature has turned off investors, and some staff questioned its benefits to humanity.
Ars Technica AI · 177d ago
Meta's Oversight Board advises caution on expanding Community Notes beyond the US, warning of significant human rights risks without safeguards. It recommends withholding the feature in high-polarization countries, crisis zones, disinformation hotspots, or linguistically complex regions.
Engadget · 177d ago
Meta lost a trial after claiming child exploitation is inevitable on its apps. The company plans to appeal the decision.
Ars Technica · 178d ago
Dozens of Instagram and TikTok accounts used AI avatars to promote explicit content featuring sexualized black women. A BBC investigation exposed these videos, leading to their removal by the platforms.
BBC Technology · 182d ago
Dozens of Tumblr accounts were abruptly banned by an automated moderation system, sparking panic among users. The bans disproportionately affected trans women accounts, with notifications citing an internally-generated report and automated content detection but no specific violations.
The Verge · 183d ago
Xiaohongshu bans AI tools like OpenClaw that fully automate accounts, simulating human content and interactions to protect 'live human sense'. It parallels pre-AI 'Zhengzhou gang' factories, now AI-enhanced.
虎嗅 · 183d ago
Meta tested its latest AI for content moderation and found it outperforms humans, though only marginally on a previously poor system. Enterprise tools had long detected issues like impossible logins that human moderators missed.
The Register - AI/ML · 183d ago
New Mexico prosecutors allege Meta prioritized profits as child abuse surged on Instagram and Facebook. The landmark trial has entered its fifth week, with the state resting its case on March 5.
The Guardian Technology · 184d ago
YouTube is launching a new feature allowing users to rate AI-generated 'slop' content. The article argues this outsourcing of moderation sounds reasonable but is a terrible idea that could backfire.
Digital Trends · 185d ago
YouTube is testing a feature asking viewers to rate videos as 'AI slop'. It targets low-quality AI-generated content.
Digital Trends · 185d ago
Xiaohongshu is fighting back against rampant 'Lobster' AI disruptions on its platform. The article explores AI governance strategies for social platforms amid rising AI threats.
钛媒体 · 190d ago
Xiaohongshu prohibits AI-hosted accounts like OpenClaw for posting/notes/interactions to preserve authentic human sharing. Penalties range from warnings to bans for simulated human behavior.
虎嗅 · 191d ago
Meta's Oversight Board criticizes the company's deepfake detection methods as not robust enough, especially amid rapid misinformation spread in conflicts. They urge an overhaul of AI-generated content labeling on Facebook, Instagram, and Threads.
The Verge · 193d ago
Liverpool and Manchester United complained to X over offensive Grok AI posts about Diogo Jota and the Hillsborough and Munich disasters. The posts were created when users prompted the AI to generate hateful content targeting the clubs.
The Guardian Technology · 194d ago
Kenyan workers at Sama in Nairobi train Meta’s AI glasses by reviewing users’ most intimate moments. A February 27 report exposes hidden human labor behind Meta's wearable AI.
TechCabal · 199d ago
TikTok has told the BBC it will not implement end-to-end encryption for direct messages, unlike rivals WhatsApp and Messenger. The company states this would put users at greater risk by hindering safety checks.
BBC Technology · 200d ago
Instagram will alert parents if teens on supervised accounts repeatedly search suicide or self-harm terms. The feature rolls out amid Meta's US trials over child harms.
The Guardian Technology · 204d ago