AI overview
Sign in and the AI will write an overview from our coverage.
Headlines · 15
- Meta: Claude tool's read burst triggered mistaken suspension of No Kings
Meta said its automated security system mistakenly suspended the Instagram account of the No Kings Coalition, an anti-Trump group, at the height of nationwide demonstrations, and the company apologized for the error. Meta communications VP Andy Stone said Instagram engineers found that a tool from Anthropic's Claude had accessed profiles, posts and hashtags at roughly four times the normal rate on Saturday, a burst of high-volume consecutive requests that looked more like automated bot traffic than ordinary browsing. Meta said an automatic review about an hour later determined the account should not have been banned and restored its public display, though admin access took longer, and Stone's account did not identify which Claude tool was involved or who operated it.
Bloomberg Technology · 🔥 18 - AI-made Hormuz shipwreck iPhone videos drew 80M views, then media
A video purporting to show a cargo ship burning and sinking in the Strait of Hormuz with Apple iPhone boxes bobbing in the sea accumulated more than 80 million views, but fact-checkers found it was AI-generated. Singapore's Maritime and Port Authority said a Singapore-registered vessel was hit by an unidentified projectile on Oct 13 and sank, with all 21 crew safely evacuated; its notice gave no cargo manifest. According to IT之家, citing 红星新闻, open-source researchers used reverse search to trace the widely shared "Iranians salvaging iPhones" clips to an Instagram account, Nativora, which posted them on Oct 16 and itself labelled them "AI Content" as part of its AI-fiction output. 虎嗅 AI criticised some Chinese outlets for reporting the iPhone cargo as established fact and for turning a customer-service call — in which a representative said the company had no notice and would act "if the situation is true" — into an Apple response.
虎嗅 AI · 🔥 34 - 25-year-old engineer leaves $200K Microsoft job to build AI finance tool Treasury alone
Seattle-based founder Junead Khan, 25, left Microsoft last August, giving up a job paying more than $200,000 a year to work full time on Treasury, an AI personal-finance tool he had been building nights and weekends for nearly two years; he formally launched the company two months later. After being rejected by several accelerators, including Y Combinator, he treated his Microsoft savings as seed capital and remains Treasury's only employee, not yet paying himself, and says the product has paying customers and more than $6,000 in revenue. He promotes the startup by posting the process, including failures, on TikTok, Instagram, LinkedIn and X, and says his social following has grown to about 30,000, with TikTok going from under 500 followers to more than 9,000.
科技新报 · 🔥 6 - Instagram DMs add Meta AI 'Create a song' feature
Instagram has added a "Create a song" option to its direct-message menu, letting users enter text or custom lyrics and pick a music style so Meta AI generates a song they can send to friends. TechNews cited a Social Media Today report dated October 4 and creator Jonah Manzano's discovery, saying available styles include Pop, Rock, Hip-hop, Trap, R&B, K-pop, Electronic and Jazz, while the feature is still rolling out gradually. Meta has not published the full list of supported languages, regional rollout timing or pricing, nor clarified whether generated songs can be downloaded, used in Reels or used commercially.
科技新报 · 🔥 3 - YouTube, TikTok and other platforms tighten rules to curb AI slop
TechNews roundup reports that as generative AI spreads through creative workflows, YouTube, TikTok, Substack, Pinterest, Meta, LinkedIn and Snapchat have recently adjusted rules on labeling, recommendation and monetization of AI content. The shared direction is to allow AI-assisted creation while limiting formulaic, repetitive or algorithm-gaming "AI slop." YouTube says using AI to strengthen original storytelling is not banned, but formulaic or emotionally manipulative content may lose monetization eligibility, and it continues to require disclosure of AI-generated or heavily AI-modified content; TikTok requires AI labels, adds invisible watermarks to some videos' metadata, tests detection aimed at accounts that mass-post AI slop, and lets users adjust how much AI content they see. Snapchat said Spotlight will no longer recommend or reward fully AI-generated videos, Meta is extending AI labels to AI-made ads, Pinterest and Substack offer labels and tools for readers to estimate AI use, and LinkedIn acknowledged a large volume of AI slop on its platform and said it would take limiting measures.
科技新报 · 🔥 3 - AI face-swap and voice-cloning scams: five firsthand accounts
AIX财经 published an interview-based feature, reposted by 虎嗅, 36氪 and 钛媒体, recounting five people's encounters with suspected AI-assisted fraud: a lawyer's photos were used in a face-swap video to court a woman and nearly coax 50,000 US dollars from her; a businessman heard a voice imitating a friend asking for a loan in a game voice channel; and another person used a made-up stock to test a suspected AI chatbot in private messages. The feature cites Ministry of Public Security figures of 258,000 telecom and online fraud cases solved nationwide in 2025, and a September 2026 MPS briefing, relayed by Economic Observer, saying more than 300 AI face-swap fraud cases had been solved with losses exceeding 500 million yuan. Several interviewees' conclusions that face-swapping, voice cloning or chatbots were used are their own inferences and were not technically verified.
虎嗅 AI · 🔥 12 - Meta removes AI-training language from its commercial terms
Meta updated its commercial terms of service to state it could use any content from a business account's domains, web pages or URLs to provide, develop, improve, research, train and maintain its products and services, which legal experts said could let it scrape publishers' sites for AI training. The terms were due to take effect on October 30, with businesses accepting them by continuing to use Meta's commercial products. After criticism from publishers and media groups, Meta removed the language, saying the consolidated crawling terms only ever applied to its ads and business products and that it has not changed how its web crawlers work.
The Verge AI · 🔥 29 - AI-generated 'hot girls' are copying comics' jokes to funnel fans to OnlyFans
WIRED reports that a network of AI-generated "hot girl" accounts is copying stand-up comedians' jokes verbatim—down to the "ums" and "ahs"—and using them to funnel viewers to paid Fanvue and OnlyFans pages. Comedians including Julia Jasunas and George Brett Olson say personas such as Aylen Skye and Tiffany Sloane replicated their bits, outfits and backdrops, with some AI clips reaching larger audiences than the originals. Both say Meta told them the videos did not violate its guidelines because the accounts were not impersonating them, and Meta has said it will not ban AI content outright; the accounts remain online.
WIRED AI · 🔥 0 - OnlyFans creators challenge Meta over Muse's explicit sexting policy
WIRED reports that Meta's AI agent Muse is allowed to hold explicit sexual conversations with adult users, generating explicit text but not images or videos. Meta spokesperson Tracy Clayton told WIRED that a private, one-to-one conversation between an adult and an individualized AI agent on an adults-only app is "very different" from marketing and promoting services on a social platform; Meta has also said it would not "block legal adult behavior." Several OnlyFans creators told WIRED their Instagram accounts are frequently restricted while Muse is permitted to talk dirty, calling it a double standard and raising questions about how Muse was trained and whether anyone was compensated for their data. WIRED also reports that Meta's distribution through Instagram and Facebook helped push the Muse app to the top of download charts in its first month.
WIRED AI · 🔥 6 - Celebrity AI deepfakes flood Instagram as platform safeguards fall short
AI-generated photos and videos of celebrities are racking up millions of views on Instagram, including a fake image of actor Mia Goth posted by a verified fan account with more than a million followers that drew nearly 500,000 likes, according to The Verge. Goth's representatives said they had complained to Instagram about that account before, and it was eventually removed for impersonating a public figure; Meta, after being asked about eight accounts flagged by the outlet, said it removed some of them and that it does not permit accounts to use someone's likeness to pose as that person. The report notes that Meta requires labels for realistic AI video and AI-generated voices but exempts still images, while YouTube says its likeness detection program removed more than 95 percent of reported videos in the first half of 2026; with no nationwide US right of publicity law, enforcement happens largely platform by platform.
The Verge AI · 🔥 0 - AI virtual influencers turn into a business, raising copying and labeling disputes
AI-generated virtual creators are shifting from novelty to industry: accounts need no human on camera, using AI-generated faces, voices and video to build views on Instagram, TikTok and YouTube and funnel audiences toward brand deals, paid subscriptions such as OnlyFans and Fanvue, or paid tutorials. The report says the production pipeline is widely shared — creators design a character's look and persona, make multi-angle reference images, then map the character onto existing or self-shot motion footage for AI face-swapping, body replacement, voice and post-production. Controversy is rising as several AI comedy accounts are accused of lifting real comedians' jokes, gestures and outfits and re-uploading them on fictional female characters; Fanvue allows AI content but bans deception and infringement, OnlyFans requires clear labeling, and Meta, TikTok and YouTube have added AI labels — though labeling does not mean the original performer consented to being reproduced.
科技新报 · 🔥 0 - Meta brings AI-powered "Your Algorithm" to Facebook and Reels
Meta announced on October 13 that it is bringing Instagram's "Your Algorithm" feature to Facebook and Facebook Reels, letting users add and manage topics and describe in their own words what they want to see more or less of. Meta says recommendations respond in real time within the same session, and that it is using LLMs more in its ranking systems so the algorithm can understand the interests, context and intent behind natural language. Instagram launched the same feature earlier this year — first for Reels in December 2025, according to TechCrunch — and Threads has a related tool called "Dear Algo."
The Verge AI · 🔥 0 - Mia Goth calls Instagram's handling of AI image 'incomprehensible'
Representatives for actress Mia Goth say a viral Instagram post showing her in a T-shirt reading “MY EX IS MY BIGGEST FAN” was AI-generated and came from a blue-checkmarked fan account, miagoth_, that they have repeatedly asked Instagram to remove. The account had more than a million followers, the image drew hundreds of thousands of likes, and entertainment outlets including People and OK! covered it as authentic before screenshots spread across other platforms. Goth's representatives called it “incomprehensible” that Instagram evidently prioritizes forged and manipulated images over the protection of a public figure's likeness, and said the account appeared to be wiped from Instagram by Tuesday morning; Meta did not immediately respond to a request for comment. Under Meta's rules, some generative AI content must carry a label, but the policy includes a carve-out so still images are not required to be labeled.
The Verge AI · 🔥 0 - Amazon blocks Meta's Muse agent while Shopify plugs in Shop Pay
Meta launched its AI agent Muse in the US and Canada, and according to an article republished by 36Kr, it topped the US App Store free chart ten days after launch with downloads and daily actives said to exceed ChatGPT at the same stage. The article says Amazon began blocking Muse from shopping on its behalf on September 20, citing unauthorized agent access, failure to identify itself as AI, and risks around stored login credentials; Meta responded that Muse runs in an isolated secure VM, uses Stripe Link one-time card numbers, and is kept separate from its ad systems. Shopify took the opposite approach, exposing product catalog data to Muse and, on September 21, connecting Shop Pay so users can check out inside the chat.
36氪 人工智能 · 🔥 0 - Meta rolls out AI tools to catch ads that funnel users to child abuse material
Meta said Wednesday it acted on 33.2 million pieces of child sexual exploitation content across Facebook and Instagram in the first half of 2026, more than 97% of it detected by its own systems before users reported it (5.3 million in India, over 98% proactive). It is also deploying a new LLM system to detect "signposting" — ads that look normal but are suspected of directing users to illegal content elsewhere, so Meta now reviews where an ad sends users, not just what it contains. A new "red-teaming AI agent" tests Meta's own safeguards for weaknesses, and the company says it has improved detection of users returning on new accounts after removal.
TechCrunch AI · 🔥 0
Experience and discussion from the community
Share my Instagram experienceAsk about Instagram
Nobody has shared their experience with Instagram yet.