OpenAI Releases a Free Tool for Detecting AI-Generated Child Abuse Images
Top Open AI AI News Stories
14 Stories
BFWAIOpen AI17 Nov 2026, 10:51 amBy BFWAI Team
OpenAI Releases a Free Tool for Detecting AI-Generated Child Abuse Images
OpenAI built an image classifier with the National Center for Missing and Exploited Children and is offering it free to vetted organizations. ChatGPT also adds warnings and 24-hour suspensions for attempts to get such material.
OpenAI built an image classifier with the National Center for Missing and Exploited Children and is offering it free to vetted organizations. ChatGPT also adds warnings and 24-hour suspensions for attempts to get such material. OpenAI released a classifier that flags suspected child sexual abuse material, including AI-generated images, and offers it free through its API to vetted organizations. It was built with NCMEC, the US clearinghouse for abuse reports. OpenAI says it flagged 48% fewer images for review at the same detection rate in tests. ChatGPT now answers attempts to seek such material with a warning, removal of the chat and a 24-hour pause. Key Capabilities:
AI signal: OpenAI says it is the first widely available classifier with a signal for whether an image may be AI-generated.
Access: Vetted organizations get it free through the API, and Tech Coalition and ROOST help widen access.
Developer guidance: OpenAI also published guidance for developers and training for investigators.
Install the Unrot app to read the full story
BFWAIOpen AI16 Nov 2026, 10:28 amBy BFWAI Team
OpenAI’s GPT-6 Astra Helps Physicist Tackle a Standard Model Problem
OpenAI says GPT-6 Astra helped physicist Nathaniel Craig develop a candidate lattice definition of the Standard Model, addressing a difficult problem in theoretical physics involving chiral fermions on discrete spacetime lattices.
OpenAI says GPT-6 Astra helped physicist Nathaniel Craig develop a candidate lattice definition of the Standard Model, addressing a difficult problem in theoretical physics involving chiral fermions on discrete spacetime lattices.
Craig supplied the overall strategy and refined the approach, while Astra helped carry the calculations to finer grids, demonstrate convergence and formalize the proof. The result is a candidate proposal rather than a settled solution, with further expert scrutiny still needed.
Key Capabilities:
AI-assisted physics: Astra contributed to mathematical work on a fundamental physics problem.
Candidate solution: The work proposes a possible lattice definition of the Standard Model.
Open for review: The underlying October preprint is publicly available for other researchers to examine.
Install the Unrot app to read the full story
BFWAIOpen AI11 Nov 2026, 06:21 amBy BFWAI Team
OpenAI Lets Businesses Claim Profiles and Build Agents in ChatGPT
Businesses in ChatGPT is a new beta where companies can claim a verified profile, update their details and see how often ChatGPT mentions them. Eligible US online businesses can also publish a Business Agent that answers customers’ questions.
Businesses in ChatGPT is a new beta where companies can claim a verified profile, update their details and see how often ChatGPT mentions them. Eligible US online businesses can also publish a Business Agent that answers customers’ questions. OpenAI launched Businesses in ChatGPT in beta on Nov 10. Through a dashboard called Business Hub, a company can claim its profile, keep its details and product feeds current, and see aggregate counts of mentions, website clicks and common topics. Eligible US ecommerce and online-service businesses can also publish a Business Agent, a labeled chat that answers shoppers from the company’s own information. OpenAI says it doesn’t share conversations with businesses. Key Capabilities:
Verification: Claim by phone, email, website domain or Shopify, with identity checks to publish changes.
Free at launch: Claiming a profile and publishing one Business Agent cost nothing at launch, per Digiday.
Ads link: Businesses can sponsor the agent as an ad destination, and OpenAI says paying doesn’t change how often ChatGPT recommends a business.
Install the Unrot app to read the full story
BFWAIOpen AI10 Nov 2026, 04:49 amBy BFWAI Team
OpenAI Lets One Dot Work Across Several of Your Computers
Dots, OpenAI’s always-on ChatGPT agents, can now connect to several local computers and run each task on the machine you choose. OpenAI says it can build and test an app on another machine while you keep coding.
Dots, OpenAI’s always-on ChatGPT agents, can now connect to several local computers and run each task on the machine you choose. OpenAI says it can build and test an app on another machine while you keep coding. OpenAI updated dots, its always-on agents in ChatGPT, so one dot can connect to several of your computers and pick one for each task. The dot’s cloud computer stays connected too. OpenAI’s examples: test an iOS change in Xcode on a second Mac and get screenshots, or run the same test on your laptop and an older-SDK machine. Before, a dot could use only one computer at a time. Key Capabilities:
Connect a computer: Open the dot’s profile in the ChatGPT app on that machine, then choose Allow under Computers.
Requirement: The ChatGPT app must be installed on every computer you connect.
Works from your phone: Access holds when you message the same dot from your phone.
Install the Unrot app to read the full story
BFWAIOpen AI06 Nov 2026, 06:49 amBy BFWAI Team
OpenAI Adds GIFs and Reusable Reference Images to ChatGPT Images 2.6
ChatGPT Images 2.6 can turn a prompt or photo into an animated GIF and remember reference images of you, pets and characters. Two new API models, Flare and Sunburst, generate images 50% faster.
ChatGPT Images 2.6 can turn a prompt or photo into an animated GIF and remember reference images of you, pets and characters. Two new API models, Flare and Sunburst, generate images 50% faster. OpenAI released ChatGPT Images 2.6 for all plans. You can turn a prompt or photo into an animated GIF and edit its frames, and save reference images of yourself, pets or characters to reuse without uploading them again. Developers get GPT-Image-2.6 Flare, which can make an image in as little as six seconds, and Sunburst, which keeps details steadier across edits. OpenAI says generation is about 50% faster. Key Capabilities:
GIFs: Adjust speed, edit individual frames, and share on repeat.
SVG and markup: Developers also get image markup and SVG export for refining images.
Wide rollout: Live in ChatGPT and Codex on web, desktop and mobile across subscription tiers.
Install the Unrot app to read the full story
BFWAIOpen AI04 Nov 2026, 09:20 amBy BFWAI Team
OpenAI Reveals Pando, a Git Backend Built for Coding Agents
Pando is a new server for Git, the tool that stores code history, built to handle the flood of commits from people and Codex agents. OpenAI says it cut median push latency by nearly 80%.
Pando is a new server for Git, the tool that stores code history, built to handle the flood of commits from people and Codex agents. OpenAI says it cut median push latency by nearly 80%. When many AI coding agents save work at once, a shared code store jams, like one checkout line for a stadium. OpenAI built Pando, a new Git backend (the server holding its code history), so pushes to different branches run in parallel. OpenAI says Pando serves nearly a petabyte of Git data a day and cut median push time by nearly 80%. It was built in 10 weeks with help from Codex agents. Key Capabilities:
Same Git for users: Engineers and agents keep using standard git commands, with no workflow change.
Why now: Commit volume has surged as Codex lets many people and agents work at once.
Not open source: OpenAI isn’t releasing Pando but is sending its Git fixes upstream.
Install the Unrot app to read the full story
BFWAIOpen AI04 Nov 2026, 09:20 amBy BFWAI Team
OpenAI Says GPT-6 Astra Helped Design a Working Hardware Device
Two OpenAI employees used GPT-6 Astra to design a device for talking to Codex in under a week, covering its enclosure, circuit board and software. OpenAI says its hardware teams’ model usage is up 23-fold since February.
Two OpenAI employees used GPT-6 Astra to design a device for talking to Codex in under a week, covering its enclosure, circuit board and software. OpenAI says its hardware teams’ model usage is up 23-fold since February. In a side project, two OpenAI employees used GPT-6 Astra to design a device for talking to Codex in under a week: its enclosure, circuit board, interface and firmware. OpenAI’s hardware teams also use Codex and ChatGPT for CAD models (3D design files), manufacturing drawings and simulations, with token usage up 23-fold since February. That measures usage, not engineering output, and human review was still needed. Key Capabilities:
Racing partner: Chip Ganassi Racing is testing turning 3D scans of parts into editable CAD files and drawings.
Mazda: OpenAI also started a vehicle-design research collaboration with Mazda.
Looking for partners: OpenAI is seeking mechanical and electrical engineering collaborators.
Install the Unrot app to read the full story
UnrotOpen AI27 Oct 2026, 10:03 amBy Unrot Team
OpenAI Launches GPT-6.1 Terra, a Cheaper Model for Computer Work
GPT-6.1 Terra understands what’s on screen, takes actions across apps and writes naturally. It costs $1 per million input tokens and $5 per million output, half the price of its predecessor.
GPT-6.1 Terra understands what’s on screen, takes actions across apps and writes naturally. It costs $1 per million input tokens and $5 per million output, half the price of its predecessor. OpenAI released GPT-6.1 Terra, a fast, lower-cost model for visual reasoning, computer use and writing. It understands what’s on your screen and takes actions across apps, built to handle high-volume work during the day. API pricing is $1 per million input tokens and $5 per million output, half of GPT-5.6 Terra and GPT-6.1 Sol. It’s in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users. Key Capabilities:
1M-token context: Handles about 1.05M tokens of context and takes images and files as input.
Matches Astra on one test: OpenAI says it equals GPT-6 Astra on Agents’ Last Exam at about a tenth of the price per token.
Also in Copilot: Available in the API as gpt-6.1-terra and in GitHub Copilot.
Install the Unrot app to read the full story
UnrotOpen AI21 Oct 2026, 04:57 amBy Unrot Team
OpenAI Lets Developers Cache Prompts for 12 Hours
OpenAI’s API now offers 12-hour prompt caching, so repeated instructions and context stay cached and cheaper even when requests are hours apart. For logged-in Codex and Work users, OpenAI’s Tibo says the cache is kept forever.
OpenAI’s API now offers 12-hour prompt caching, so repeated instructions and context stay cached and cheaper even when requests are hours apart. For logged-in Codex and Work users, OpenAI’s Tibo says the cache is kept forever. Every time an AI app resends the same long instructions, it pays to process them again. Prompt caching works like a waiter remembering your usual order. OpenAI’s API now lets developers keep that memory for 12 hours, instead of the default 30 minutes. Cached input costs less, which suits agents that run all day. For logged-in Codex and Work users, OpenAI’s Tibo says the cache is kept forever. Key Capabilities:
Pick a timer: Choose 12 hours, keep the 30-minute default, or mix both in one request.
Lower rate: Cached input is billed at the cheaper cached-input price, even when requests are hours apart.
Install the Unrot app to read the full story
UnrotOpen AI17 Oct 2026, 03:33 amBy Unrot Team
OpenAI Makes Codex and Work Go Further on the Same Usage
OpenAI says Codex and ChatGPT Work now give more effective usage by reusing context across subagents, caching context longer, and routing requests to get more cache hits.
OpenAI says Codex and ChatGPT Work now give more effective usage by reusing context across subagents, caching context longer, and routing requests to get more cache hits. OpenAI made Codex and Work stretch your usage further. When AI agents work, they keep re-reading the same background, like handing every helper a fresh copy of the manual. Now helper agents (subagents) share that context, saved context stays cached longer, and smarter request routing sends requests to servers that already hold it, called cache hits. Same plan, more work done. Key Capabilities:
Shared context: Subagents reuse context instead of each starting from scratch.
Longer cache: Saved context is kept for longer, so more requests skip repeated work.
Day 12 of 28: Part of Tibo’s 28-day series of daily Codex and Work improvements.
Install the Unrot app to read the full story
UnrotOpen AI17 Oct 2026, 03:32 amBy Unrot Team
ChatGPT’s Chrome Extension Can Now Read and Control Your Tabs
In Chat mode, ChatGPT can look across your open Chrome tabs to answer questions. In Work mode, it can fill forms, send messages, download files and group tabs.
In Chat mode, ChatGPT can look across your open Chrome tabs to answer questions. In Work mode, it can fill forms, send messages, download files and group tabs. OpenAI updated the ChatGPT extension for Chrome so it can use the tabs you already have open. In Chat, it compares products, summarizes research or pulls details across several pages, with no copy and paste. In Work, ChatGPT can control tabs to send messages, fill forms, download files or group cluttered tabs. OpenAI says it asks permission before using browser history or handling files. Key Capabilities:
Chat mode: Reads across open tabs to compare, summarize and answer questions.
Work mode: Controls tabs to send messages, fill forms, download files and group tabs.
Beyond Chrome: Also runs on Edge, Brave and Opera, with a side-by-side panel.
Install the Unrot app to read the full story
UnrotOpen AI17 Oct 2026, 03:32 amBy Unrot Team
ChatGPT Pro Gets Faster GPT-6 Pro and Parallel Agents
OpenAI upgraded GPT-6 Pro in ChatGPT with faster responses and better use of plugins and tools. Multiple agents can now work on different parts of a task at once, in the Chat tab for Pro users.
OpenAI upgraded GPT-6 Pro in ChatGPT with faster responses and better use of plugins and tools. Multiple agents can now work on different parts of a task at once, in the Chat tab for Pro users. OpenAI upgraded GPT-6 Pro in ChatGPT for Pro subscribers. It now answers faster, uses plugins and tools better, and can hand parts of a big task to several agents that work at the same time. Users can follow each agent’s progress and open finished tasks to review the work. It’s like a manager watching a team split a project, and it appears in the Chat tab. OpenAI gave no speed figures or examples. Key Capabilities:
Parallel agents: Several agents tackle different parts of one task at the same time.
Visible progress: You can follow each agent and open completed tasks to check their work.
Pro users only: Available to eligible Pro users in the Chat tab, rolling out now.
Install the Unrot app to read the full story
UnrotOpen AI16 Oct 2026, 02:51 amBy Unrot Team
OpenAI Updates ChatGPT Voice With GPT-Live-1.1 and 13 New Voices
GPT-Live-1.1 adds new voices and more natural accents in six languages, and says “checking...” less while searching. It’s rolling out to all users in ChatGPT, Codex and dots.
GPT-Live-1.1 adds new voices and more natural accents in six languages, and says “checking...” less while searching. It’s rolling out to all users in ChatGPT, Codex and dots. OpenAI released GPT-Live-1.1, an update to the voice model behind ChatGPT Voice, Codex and dots. It adds 13 new voices across six languages plus English, and improves accents in many non-English languages. It also says “checking...” less often while searching. The update is live for all users now, and OpenAI says API access is coming soon. Key Capabilities:
13 new voices: Two each in French, German, Hindi, Japanese, Korean and Mexican Spanish, plus an English voice called Bubbie.
Better accents: OpenAI says accents and multilingual support are considerably better.
Where it works: Voice in ChatGPT, Codex and dots, with API access still to come.
Install the Unrot app to read the full story
UnrotOpen AI14 Oct 2026, 03:09 amBy Unrot Team
FBI Arrests OpenAI Employee Over Hidden Recording Device
OpenAI says it found a sophisticated recording device in an executive conference room and traced it to one employee using its own models. The employee was fired and is accused of planting it.
OpenAI says it found a sophisticated recording device in an executive conference room and traced it to one employee using its own models. The employee was fired and is accused of planting it. OpenAI says it found a sophisticated recording device hidden in an executive conference room at its San Francisco office in late September. The FBI arrested an employee on Oct 9, charged with intercepting oral communications. OpenAI fired him. OpenAI says it used its own models to identify the device and who planted it, and has no sign that model weights, source code or customer data were accessed. The FBI investigation is ongoing. Key Capabilities:
Allegation only: Prosecutors allege the employee planted it, and he is charged with intercepting oral communications.
Models used: OpenAI says it used its own models to trace the device back to the employee.
Second device: Prosecutors say a similar device was found in the employee’s unattended bag.
Install the Unrot app to read the full story
Explore Topics
Explore Other AI News Categories & Companies
Discover news stories, research breakthroughs, and tool updates across other AI categories.