--- title: Hindsight puts up to $1M behind its AI content... description: Google's Perspective API shuts down on December 31, forcing platforms to find a new toxicity-detection tool. Hindsight is pitching itself as the replacement,... canonical: https://daily.dev/posts/hindsight-puts-up-to-1m-behind-its-ai-content-moderation-as-google-shuts-down-perspective-api-lsniasfkh twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Hindsight puts up to $1M behind its AI content moderation as Google shuts down Perspective API | daily.dev og:description: Google's Perspective API shuts down on December 31, forcing platforms to find a new toxicity-detection tool. Hindsight is pitching itself as the replacement,... og:url: https://daily.dev/posts/hindsight-puts-up-to-1m-behind-its-ai-content-moderation-as-google-shuts-down-perspective-api-lsniasfkh og:image: https://api.daily.dev/og/posts/lsniASfkh.png og:image:alt: Hindsight puts up to $1M behind its AI content moderation as Google shuts down Perspective API og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Hindsight puts up to $1M behind its AI content moderation as Google shuts down Perspective API **[The Next Web](https://daily.dev/sources/tnw)** · 6 min read · 0 upvotes · 0 comments ## Summary Google's Perspective API shuts down on December 31, forcing platforms to find a new toxicity-detection tool. Hindsight is pitching itself as the replacement, claiming its model catches 'coded language' like numeric codes and swapped characters that Perspective misses. In benchmarks against 2,205 posts, Hindsight says it detected 97.9% of coded-language cases versus 65% or below for 9 competing models including Mistral, OpenAI, Meta, Azure and AWS. It uses three specialized sub-600M-parameter models instead of one large model, and offers clients up to $1 million if a post it labels 'Safe' causes a documented liability event. The launch cohort is capped at 4 commercial clients. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### When is Google shutting down the Perspective API and what should I use instead? Google's Jigsaw unit shuts down Perspective API on December 31, ending its toxicity-scoring service. Any moderation pipeline calling it will fail starting January 1. One proposed replacement is Hindsight, a startup claiming 97.9% detection of coded hate speech (numeric codes, extremist slang, l33tspeak, phonetic misspellings) versus 65% or below for 9 other tested models. _Teams racing to replace Perspective before the cutoff can track moderation tool comparisons on daily.dev._ ### How does Hindsight's content moderation model differ from using one large general-purpose AI model? Hindsight uses three specialized base models, each under 600 million parameters, rather than a single large model, which it says outperforms models up to 14 times its size on coded-language detection. It was trained on 83 MB of real online posts from 18 datasets, including 292,000 human-labeled posts, across 164 trained variants, with the best performer promoted as the 'Champion'. _Developers weighing specialized versus general-purpose models for a task can follow this trade-off on daily.dev._ ### What is 'coded language' in online content moderation and why is it hard to detect? Coded language refers to hate speech disguised to look harmless, such as numeric codes like '1488', extremist slang like 'boogaloo', swapped characters ('l33tspe@k'), and phonetic misspellings of slurs like 'phag'. Standard toxicity scorers like Perspective API judge how toxic text 'sounds' on the surface, so they catch obvious insults but miss language deliberately designed not to sound toxic. _Anyone building moderation systems that must catch disguised hate speech can dig into this challenge on daily.dev._ ## Similar posts on daily.dev - [Meta’s latest AI improves its terrible content moderation](https://daily.dev/posts/meta-s-latest-ai-improves-its-terrible-content-moderation-y6ptsdngz) · The Register · 0 upvotes · 0 comments - [Probably raises $9M to build a more reliable kind of AI](https://daily.dev/posts/probably-raises-9m-to-build-a-more-reliable-kind-of-ai-mpymluh3i) · TechCrunch · 2 upvotes · 0 comments - [Weekly Dose \#22 - AI Is Getting Cheaper. Your Model Choices Just Got More Interesting.](https://daily.dev/posts/weekly-dose-22---ai-is-getting-cheaper-your-model-choices-just-got-more-interesting--aqw4e4g4p) · Machine Learning Pills · 0 upvotes · 0 comments --- Tags: [#nlp](https://daily.dev/tags/nlp), [#ai-safety](https://daily.dev/tags/ai-safety) [View this post on daily.dev](https://daily.dev/posts/hindsight-puts-up-to-1m-behind-its-ai-content-moderation-as-google-shuts-down-perspective-api-lsniasfkh) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Hindsight puts up to $1M behind its AI content moderation as Google shuts down Perspective API","url":"https://daily.dev/posts/hindsight-puts-up-to-1m-behind-its-ai-content-moderation-as-google-shuts-down-perspective-api-lsniasfkh","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/hindsight-puts-up-to-1m-behind-its-ai-content-moderation-as-google-shuts-down-perspective-api-lsniasfkh"},"datePublished":"2026-10-23T18:52:42.547Z","dateModified":"2026-10-23T18:53:13.129Z","description":"Google's Perspective API shuts down on December 31, forcing platforms to find a new toxicity-detection tool. Hindsight is pitching itself as the replacement,...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/718cee734863455172ebb326cf36abb3?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/718cee734863455172ebb326cf36abb3?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"The Next Web","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"The Next Web","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/tnw","url":"https://daily.dev/sources/tnw"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/hindsight-puts-up-to-1m-behind-its-ai-content-moderation-as-google-shuts-down-perspective-api-lsniasfkh","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"nlp,ai-safety","timeRequired":"PT6M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"The Next Web","item":"https://daily.dev/sources/tnw"},{"@type":"ListItem","position":3,"name":"Hindsight puts up to $1M behind its AI content moderation as Google shuts down Perspective API"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/hindsight-puts-up-to-1m-behind-its-ai-content-moderation-as-google-shuts-down-perspective-api-lsniasfkh#faq","mainEntity":[{"@type":"Question","name":"When is Google shutting down the Perspective API and what should I use instead?","acceptedAnswer":{"@type":"Answer","text":"Google's Jigsaw unit shuts down Perspective API on December 31, ending its toxicity-scoring service. Any moderation pipeline calling it will fail starting January 1. One proposed replacement is Hindsight, a startup claiming 97.9% detection of coded hate speech (numeric codes, extremist slang, l33tspeak, phonetic misspellings) versus 65% or below for 9 other tested models. Teams racing to replace Perspective before the cutoff can track moderation tool comparisons on daily.dev."}},{"@type":"Question","name":"How does Hindsight's content moderation model differ from using one large general-purpose AI model?","acceptedAnswer":{"@type":"Answer","text":"Hindsight uses three specialized base models, each under 600 million parameters, rather than a single large model, which it says outperforms models up to 14 times its size on coded-language detection. It was trained on 83 MB of real online posts from 18 datasets, including 292,000 human-labeled posts, across 164 trained variants, with the best performer promoted as the 'Champion'. Developers weighing specialized versus general-purpose models for a task can follow this trade-off on daily.dev."}},{"@type":"Question","name":"What is 'coded language' in online content moderation and why is it hard to detect?","acceptedAnswer":{"@type":"Answer","text":"Coded language refers to hate speech disguised to look harmless, such as numeric codes like '1488', extremist slang like 'boogaloo', swapped characters ('l33tspe@k'), and phonetic misspellings of slurs like 'phag'. Standard toxicity scorers like Perspective API judge how toxic text 'sounds' on the surface, so they catch obvious insults but miss language deliberately designed not to sound toxic. Anyone building moderation systems that must catch disguised hate speech can dig into this challenge on daily.dev."}}]} ```