Offering an Adult Mode on ChatGPT was a promise made by CEO Sam Altman in October 2025, but quickly ran into problems, leading to delays, and later postponement.
Continue reading “OpenAI scraps Adult Mode «indefinitely,» Financial Times reports”
Tag: openai
OpenAI axes Sora video app and API, and it won’t live on in ChatGPT
The video generation app had amassed 920 million users since December 2025 and was for a while the number one app on the App Store, before declining to #165 recently.
Closing the free app, estimated to cost $15 million per day, opens up resources for OpenAI’s recent focus on coding and business, as it was labelled a «side quest» internally.
With Sora discontinued, OpenAI is also leaving behind a $1 billion deal with Disney — which had licensed some of its characters for use on the platform. Disney says they are open to new investments, and «respects» OpenAI’s decision.
Read more: The Wall Street Journal, Reuters and Tibor Blaho.
As OpenAI prepares to show ads to all Free and Go users, advertisers are giddy
That means the test with showing some ads to about 5% of users is coming to an end, and the full plan will start up just after easter.
The limited advertising has so far been a success. The main complaint from advertisers is that it’s going too slow, according to CNBC. Most of them are happy and ready to spend more — with more varied ads.
— We’re encouraged by early signals from users and participating brands, and continue to see strong interest from advertisers, OpenAI tells CNBC.
The advertising program on Free and Go tiers is expected to earn OpenAI about $1 billion per year, and usher in a third tier for advertisers in addition to Search, Social, and Retail.
Read more: The Information (paywalled), Reuters, and CNBC.
Labs are hiring experts to protect against «catastrophic misuse»
The role is for someone with long, PhD level experience in «chemical weapons and/or explosives defence,» the LinkedIn post says.
It would be helpful if the person has an «understanding of radiological materials,» the posting goes on, and says the candidate will be «tackling critical problems in preventing catastrophic misuse.»
OpenAI is not far behind in worrying about these issues, and also has a job post open for much the same, but they are looking for someone with machine learning experience from red-teaming in order to safeguard their AI’s responses.
Using any AI for developing these kinds of weapons is of course against all the labs’ terms of use, but as the models grow more capable, they also need more safeguards.
Read more: Anthropic’s job post, OpenAI’s job post, writeups on the BBC and Mashable.
OpenAI plans to combine Codex, ChatGPT and Atlas in «super app»
The app will make it easier for teams within OpenAI to work together, the WSJ reports, and will help other users with productivity-related tasks, as they double down on enterprise users.
The standalone ChatGPT app will not be affected by the move, although the paper notes that OpenAI feels it has lost attention by focusing on «side quests» like the Sora app — now rumored to get included in ChatGPT proper.
OpenAI’s Fidji Simo will be leading the super app effort, and she tweets that:
— When new bets start to work, like we’re seeing now with Codex, it’s very important to double down on them and avoid distractions.
Read more: The Wall Street Journal and CNBC.
Codex grows to 2 million weekly users, acquires Python developers Astral
Some of the most beloved and, importantly, used Python developer tools come from the company, which will now be supported by OpenAI.
The deal for roughly 32 employees will strengthen Codex by integrating the tools that have «hundreds of millions of downloads per month,» according to Astral themselves.
OpenAI will continue to maintain the open source projects, and by gaining access to them — and the engineers’ knowhow — for Codex’s AI agents, they will be able to work more closely with the tools.
Read more: OpenAI’s announcement, Astral’s announcement, and CNBC.
OpenAI upgrades GPT-5.3-instant to be «less clickbait-y» in its responses
But many had noticed that it had become filled with follow-up questions for simple queries, offering «one strange trick,» «would you like me to tell you three things that…» and «You’ll never believe…»
These teaser-style responses were not just annoying, but sometimes frustrating — as if the bot had become optimized for engagement and tried to keep the conversation going after already answering the query.
The good news is that as of March 16, 2026, OpenAI has upgraded the model to show less of this slop, and users should already be noticing an improvement in «follow-up tone.»
Read more: OpenAI’s update page, Android Headlines.
ChatGPT’s «adult mode» hotly debated at OpenAI, will be smutty, but not porn
Touted by CEO Sam Altman as letting «adults be adults» in October 2025, it was later delayed and then deprioritized last week.
It now seems the company’s internal advisory board is against going forward with the feature, saying it could foster «unhealthy emotional dependence,» Mashable writes.
Also holding back the launch is the fact that ChatGPT’s age checks aren’t that good, and has a rather large error rate of 12% on identifying kids and teens, The Verge reports.
100 million under 18s use ChatGPT every week, which would mean that some 12 million of them could be classified as adults and exposed to «sexualized conversations.»
The feature is currently postponed due to «other priorities,» but is said to skirt images, voice and video for pure text, and will supposedly be «smutty,» not «pornographic,» the Verge says.
Read more: The Wall Street Journal (paywalled), Mashable and The Verge.
OpenAI launches GPT-5.4 mini and nano for Free and Go tiers
GPT-5.4-mini can be accessed today by choosing «Thinking» from the model picker on the Free and Pro tiers.
The model is stronger than 5-mini on reasoning, multimodal understanding, tool use and subagents, OpenAI says.
It is also available in Codex, where it uses only 30% of tokens for the same task compared to 5.4 proper — and is usable for workloads where latency and speed are important.
Also introduced today is GPT-5.4-nano, which is for the times when speed and cost matter most. It costs only $0.20/$1.25 per 1M tokens, and is handy for mass classification, ranking, coding subagents and compaction.
Nano doesn’t support web search or computer use, and is only available in the API.
Read more: OpenAI’s introduction, Tibor Blaho on X.com, Engadget.
Encyclopedia Britannica sues OpenAI over copyright and trademarks
This erodes their copyright, they claim, and «starves publishers of revenue,» because people get their answers directly in ChatGPT rather than getting referred to their websites, TechCrunch writes.
Britannica is also claiming that ChatGPT erodes its trademarks by producing hallucinations and attributing them to their products.
There have been a lot of copyright lawsuits since the early days of LLM chatbots, but only a few have been settled. One judgment is from the Anthropic case, where a judge found that training on copyrighted works was «no different than training schoolchildren to write well».
Other judges have been more skeptical.
Read more: Reuters, TechCrunch, Engadget.
Sora 2 video generation could be heading for ChatGPT «soon»
According to The Information’s x.com post, the release seems imminent, while Blaho has spotted Sora generation mentions and ID’s in the latest Andoid beta ChatGPT App.
Sora 2 was launched as a quasi social network with its own app in October 2025 and was invite-only.
Judging by early search interest on Google for «Sora invite code» soon after launch, interest for the app greatly exceeded capacity and available invites.
It produces 10-second clips from prompts or pictures and promises to put your likeness in anything you can imagine. It briefly went viral, running up some 920 million weekly users before dropping from #1 to #165 in the App Store.
Putting the app’s features into the mainline GPT app will bring it up to par with Google’s AI subscriptions, which gives users access to the video generator Veo 3 on the $25 Pro plan.
Read more: The Information (paywalled), Tibor Blaho on X, and Reuters.
OpenAI puts «adult mode» on the back burner, for now
It was supposed to debut in the first quarter of 2026, after first being conceived as «letting adults be adults» in October 2025. It was supposed to run in tandem with age verification, rolled out this January.
OpenAI says that «we still believe in the principle of treating adults like adults, but getting the experience right will take more time,» writes Axios.
The postponement is due to the company focusing on other high priority issues, such as «intelligence, personality improvements, personalization,» and «making the experience more proactive.»
Read more: Axios and TechCrunch.
OpenAI’s robotics and hardware head leaves in Pentagon protest
I resigned from OpenAI. I care deeply about the Robotics team and the work we built together. This wasn’t an easy call. AI has an important role in national security. But surveillance of Americans without judicial oversight and lethal autonomy without human authorization are…
— Caitlin Kalinowski (@kalinowski007) March 7, 2026
Caitlin Kalinowski, head of robotics and consumer hardware at OpenAI, makes no qualms about her feelings on their new contract with the Pentagon, and is resigning in protest, saying that:
— Surveillance of Americans without judicial oversight and lethal autonomy without human authorization are lines that deserved more deliberation than they got.
Continue reading “OpenAI’s robotics and hardware head leaves in Pentagon protest”
OpenAI’s new ChatGPT-5.4 has native computer use and less hallucinations
— Together with advances in general reasoning, coding, and professional knowledge work, GPT‑5.4 enables more reliable agents, faster developer workflows, and higher-quality outputs across ChatGPT, the API, and Codex, OpenAI writes.
On hallucinations, it is 33% less likely to be wrong in its responses, and 18% less apt to have mistakes in replies compared to GPT‑5.2.
Continue reading “OpenAI’s new ChatGPT-5.4 has native computer use and less hallucinations”
OpenAI ships Codex app for Windows — running in the Linux subsystem
The official Codex app has finally arrived for Windows, almost a month to the day after first debuting on macOS.
It runs natively in the Windows Subsystem for Linux, and has integrated terminals for PowerShell, Command Prompt, and Git Bash.
It also has all the «regular» features the Mac version has — and is a fully integrated multi-agent coding environment.
The Windows app is also properly sandboxed, so you can block it from accessing files outside your working folder and «prevent outbound links» unless you approve it, writes OpenAI’s Andrew Ambrosino on x.com.
In addition, your session history is stored on your OpenAI account, making it possible to start coding on a Mac and finish it on Windows without losing work, Engadget notes.
The app is available today from the Microsoft Store.
Read more: Andrew Ambrosino’s announcement, Engadget and Microsoft Store.