惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

博客园_首页
阮一峰的网络日志
阮一峰的网络日志
S
Secure Thoughts
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
C
Cyber Attacks, Cyber Crime and Cyber Security
T
Threat Research - Cisco Blogs
P
Privacy & Cybersecurity Law Blog
The Hacker News
The Hacker News
H
Heimdal Security Blog
W
WeLiveSecurity
L
LINUX DO - 热门话题
Hacker News: Ask HN
Hacker News: Ask HN
WordPress大学
WordPress大学
The Last Watchdog
The Last Watchdog
Hugging Face - Blog
Hugging Face - Blog
博客园 - 【当耐特】
D
DataBreaches.Net
I
Intezer
Webroot Blog
Webroot Blog
C
Cisco Blogs
AWS News Blog
AWS News Blog
博客园 - 聂微东
T
The Blog of Author Tim Ferriss
V
Vulnerabilities – Threatpost
罗磊的独立博客
Google DeepMind News
Google DeepMind News
N
Netflix TechBlog - Medium
Schneier on Security
Schneier on Security
宝玉的分享
宝玉的分享
博客园 - 叶小钗
PCI Perspectives
PCI Perspectives
D
Docker
Scott Helme
Scott Helme
NISL@THU
NISL@THU
J
Java Code Geeks
B
Blog RSS Feed
Google Online Security Blog
Google Online Security Blog
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
T
The Exploit Database - CXSecurity.com
AI
AI
美团技术团队
Cloudbric
Cloudbric
月光博客
月光博客
P
Proofpoint News Feed
T
Tailwind CSS Blog
Google DeepMind News
Google DeepMind News
小众软件
小众软件
www.infosecurity-magazine.com
www.infosecurity-magazine.com
The Cloudflare Blog
cs.CV updates on arXiv.org
cs.CV updates on arXiv.org

Latest from TechRadar in Chatgpt

OpenAI replaced GPT-4o, but some users still refuse to let it die — now they’re celebrating its ‘birthday’ in Times Square with a video billboard ‘He needed to have total control over it’ — Altman testifies Musk never trusted shared leadership… I tried a viral 'backwards calendar' ChatGPT prompt — and it completely changed how I plan my week OpenAI snaps up consulting company to help spread the word about AI I asked ChatGPT and Gemini how to make French Toast as good as my mother used to make — one nailed the… OpenAI's ‘Trusted Contact’ feature for ChatGPT users in crisis is a sign AI is becoming something much… ChatGPT now lets you nominate a Trusted Contact who gets alerted if your interaction with AI 'indicates a serious… OpenAI has 3 new AI voice models that the ChatGPT maker says will ‘unlock a new class of voice apps for… AI may kill the app grid, but I still think complex tasks need apps — and that I asked ChatGPT to ruin my photos with ugly 90s-style MS Paint art — and the results are weirdly brilliant Google is bringing Gemini to Mac in a bid to help organize your files and much more ChatGPT just gave me a glimpse of my life in three years ‘Without me, OpenAI wouldn’t exist,’ says Elon Musk as courtroom clash with Sam Altman turns personal — and exposes a deeper fight over who really built the company behind ChatGPT Sam Altman says some companies are ‘AI washing’ by blaming unrelated layoffs on the technology — but… I used ChatGPT Images 2.0 to meet my childhood self — and this nostalgic photo prompt is going viral for a reason ‘The worst-case situation is where it is a Terminator situation’ — Elon Musk invokes killer robots in… Gen Z hate AI? The Musk vs Altman trial heats up, OpenAI phone rumors buzz and more of the week’s most surprising… OpenAI is making ChatGPT accounts much more secure – including some literal physical security keys I asked ChatGPT to reimagine The Devil Wears Prada 2 ending based on the shocking Runway magazine AI twist in the sequel — and the results aren't as dreadful as you'd think Everyone’s switching from ChatGPT to Claude — but new tests say neither is the smartest free AI, and the… ChatGPT just made it easier to pick the right model, just like Gemini does — here’s when to use Instant,… I noticed ChatGPT slowly drifting off topic in long chats — this tiny prompt forces it to reset itself every few messages and keeps the conversation surprisingly on track Sam Altman just dropped a big hint that GPT-6 is coming soon — ‘with extra goblins’ 'I won’t provide instructions, tactics, or advice that could help someone commit a crime': ChatGPT claims it won't assist would-be felons, despite claims to the contrary from Florida AG ChatGPT just announced it can finally pass the simple ‘how many “r”s in strawberry’ test, but users are still tripping it up by switching to ‘cranberry’ Musk vs Altman heads to trial in a battle that could reshape the future of AI for everyone I swapped my iPhone for a pair of Ray-Ban Meta (2nd gen) on vacation, and I felt liberated — but these smart glasses have a way to go yet I stopped asking AI for answers and started asking for frameworks — and suddenly it all clicked Would you buy a ChatGPT-powered iPhone rival? OpenAI is reportedly developing a smartphone chip, which teases the… I compared ChatGPT Images 2.0 and Google’s Nano Banana 2 using real-world prompts — from portraits to product shots — and the AI image generator that came out on top genuinely surprised me I tried using ChatGPT to follow The 7 Habits of Highly Effective People step by step — and it made one of the most famous self-improvement systems feel surprisingly simple to put into practice AI did what? ChatGPT's big upgrade, Google’s plans to make AI invisible, DeepSeek’s return and more of the week’s most surprising developments 'We love you, and we want you to win'  — OpenAI releases GPT-5.5 for ChatGPT 'Just use the manual': I followed ChatGPT's PC-building instructions, and all I got was a huge headache… ChatGPT Images 2 made me a perfect magazine — then I realized I couldn’t do anything with it ChatGPT can threaten to ‘key your car’ and become increasingly abusive if you prompt it just right, new… Apple is about to reinvent Siri at WWDC this year — and it might finally feel like ChatGPT 'Not just generating images. It’s thinking'  — ChatGPT Images 2.0 could fundamentally change… This browser plugin lets you track your AI 'water footprint' — and ‘pay it back’ to… 'Building a PC in 2026 is an absolute treat': I asked ChatGPT and Gemini to pick my perfect gaming rig, and it… AI did what? Gemini, Val Kilmer, Claude Mythos and more of the week’s most surprising developments ChatGPT was down for many — as OpenAI says it's 'monitoring the recovery' I tried a new ChatGPT tool to help me find a better wedding venue — and I’m very glad it isn’t… I used ChatGPT to envision my kid’s doodles as real animals, and they looked surprisingly lifelike OpenAI reveals its Mythos rival designed for cybersecurity pros OpenAI says Microsoft has ‘limited our ability’ to build customer base, praises new Amazon relationship This simple ‘assignment’ prompt flips ChatGPT’s biggest weakness — and I wish I’d used it… Sam Altman’s weekend: second home attack, a furious blog post, and a growing backlash 'A secret agreement': Microsoft urges judge to throw out claims it colluded with OpenAI to boost ChatGPT… Microsoft says Copilot is for ‘entertainment' not work, Meta’s Muse Spark and 7 other AI stories you… ‘I hadn’t verified a single thing’: Using ChatGPT for Iran war news changed how I trust information I matched the upgraded Meta AI against ChatGPT, and you can really tell which AI has social media roots ChatGPT’s hidden backup model just got smarter — as OpenAI adds a cheaper Pro option Could Google Gemini surpass ChatGPT as the biggest AI chatbot within a year? If it maintains its current growth… I tried using ChatGPT in CarPlay — and it was weirder than I expected OpenAI says AI could mean a 4-day week ‘with no loss in pay’ — but warns jobs and wealth are at risk ‘Find a way to work together’ — Sam Altman’s message to the Department of Defense and Anthropic 'Not the budget smart glasses I've been waiting for' — Rokid's AI Glasses aren't the Meta Ray-Ban rivals I hoped they would be Google has doubled its storage allowance for its AI Pro plan, and it’s just one-upped ChatGPT — here’s… OpenAI doesn’t just want to answer your questions — it wants to run your digital life I asked ChatGPT how to feel less exhausted — what it said wasn’t new, but it worked ChatGPT comes to Apple CarPlay but only if you are willing to talk to a robot 'We are growing revenue four times faster than the companies who defined the Internet and mobile eras': OpenAI says it's making $2 billion a month — mostly from enterprise users Why you shouldn’t ask ChatGPT for relationship advice — it’ll just tell you you’re right and 'may worsen rather than resolve conflict' I use these 5 simple ‘ChatGPT codes’ every day — and they instantly improve my results I used a simple ChatGPT ‘gig finder’ prompt — and it found bands playing near me I’d never heard… 'A hard truth for the AI era: don’t assume AI tools are secure by default': OpenAI patches flaw allowing silent data leakage from ChatGPT conversations without users ever knowing 'Not just development tools': Security experts discover critical flaw in OpenAI's Codex which could compromise entire enterprise organizations I turned on Location in ChatGPT — and it changed how I searched I connected my Dropbox to ChatGPT — and it changed how I find everything 'Sum up what you know about me' — I tried Gemini’s new memory import and it fixes AI’s most… Claude is limiting usage more aggressively during peak hours — here’s what changed ‘This was never just about sex’ — ChatGPT’s ‘adult mode’ being shelved reveals a… AI is making us all sound the same at work — I tested it to see if it’s true ChatGPT just made its shopping tools more visual — here are 5 prompts you can try right now for better results I write about AI for a living — what people confessed to me about using ChatGPT surprised me I felt weird using ChatGPT in public so I tried this extension that disguises it as a Google Doc Shopping in ChatGPT is about to get a lot more interactive – users promised 'improved speed, relevance and… I asked ChatGPT to grade my interview answers — it was more useful than real interviews I use the 'invert' prompt to answer problems before they arise “Thank you for getting us to this point” — I thought Sam Altman’s tweet about coders was nice… ChatGPT might not be using the model you think — and it’s also hiding others in settings OpenAI is making an all-in-one 'superapp' that combines Codex, ChatGPT, and Atlas browser for maximum… Here’s how I write to make sure nobody thinks I’m an AI AI is helping UK SME workers save 5.2 hours a week, but some businesses are really struggling to keep up I asked ChatGPT and Gemini to make me a March Madness bracket — only one blew me away I told ChatGPT there was an 'extremely lazy person' here — and its answers got way better I asked ChatGPT what to watch across 6 streaming apps — and it actually nailed it I tested the new ChatGPT-5.4 mini and nano models — and I didn’t expect them to be this powerful This simple ChatGPT trick forces the AI to poke holes in its own logic 'Its most distinguishing characteristic is its humanity' one ChatGPT user says — and Sam Altman agrees as OpenAI takes steps to reduce ‘teaser-style phrasing’ in responses I tested the ‘future self’ prompt in ChatGPT and couldn’t believe how personal the advice it gave me… I tried ChatGPT’s new visual explanations for math — and I was surprised how intuitive it feels I got tired of ChatGPT trying to bait me with follow-up questions after every answer — so I made this one easy… ‘There’s no going back’ — still thinking of quitting ChatGPT? Make sure you do these 3 things… I tried the Shazam app for ChatGPT – and it’s a surprisingly natural fit for the chatbot ChatGPT users are flocking to Claude then realizing they can’t use it the same way — ‘For new users,… OpenAI releases Codex Security to spot the next big cyber risks to your company, promises to 'identify complex vulnerabilities that other agentic tools miss' Sam Altman says GPT-5.4 is his 'favorite model to talk to' — but admits OpenAI still needs to fix these… ChatGPT has delayed the roll out of its adult mode again: 'Getting the experience right will take more time'
ChatGPT just got a major personality overhaul — fewer hallucinations, fewer emojis, much shorter answers, and I can already notice the difference
https://www.techradar.com/sg/author/graham-barlow · 2026-05-06 · via Latest from TechRadar in Chatgpt
ChatGPT-5.5 logo on a smartphone.
(Image credit: Getty Images / CFOTO )

  • OpenAI says GPT-5.5 Instant cuts hallucinations by 52.5%
  • ChatGPT responses are now shorter, cleaner, and less overformatted
  • New memory sources show exactly what ChatGPT knows about you

You think you know ChatGPT, right? It can be terribly sycophantic, produce endless bullet-point lists in response to your questions, and sometimes it even goes overboard with annoying emojis. Worst of all, it occasionally hallucinates — essentially making things up.

Well, OpenAI is trying to change all of that overnight.

GPT-5.5 Instant is now rolling out to all ChatGPT users everywhere, including Free users. The previous GPT-5.3 Instant model will remain available for paid users through the Configure menu for the next three months.

The headline features of GPT-5.5 Instant include 52.5% fewer hallucinated claims on high-stakes prompts across areas like medicine, law, and finance, alongside stronger performance in image reasoning, science, and math. Responses are also designed to be clearer, shorter, and less overformatted.

OpenAI also says GPT-5.5 Instant has “a more natural conversational tone, and better use of the context you’ve already shared when personalization can help”.

That means ChatGPT is now better at using relevant context from past chats, uploaded files, and connected Gmail accounts.

Managing memory sources

New in GPT-5.5 Instant are memory sources. These give you more visibility into the context that helped personalize a response and let you manage or delete that memory source if it’s providing incorrect information.

Sign up for breaking news, reviews, opinion, top tech deals, and more.

When a response is personalized, you can now click the ‘Sources’ option at the end of the response to see what context was used. That could include saved memories, previous chats, or connected services.

You can then delete or correct outdated information using the three-dot menu next to the source.

Managing memory sources in ChatGPT-5.5 Instant

(Image credit: OpenAI)

So, if ChatGPT has incorrect information about where you live, for example, you can click the menu button and select ‘Make a correction’ to update it.

Memory sources are rolling out across all ChatGPT consumer plans on the web and coming soon to mobile. OpenAI says availability of specific personalization sources may vary by region.

What it’s like to use GPT-5.5 Instant?

I’ve been using GPT-5.5 Instant today, and the first thing I noticed was how concise it felt compared to the previous version I’d been using, GPT-5.3 Instant. And yes, that means OpenAI has skipped GPT-5.4 Instant entirely in its jump to GPT-5.5.

Ask GPT-5.5 Instant for something, and it usually just gives you exactly what you asked for without anything superfluous. It no longer feels desperate to impress you every few sentences.

It also doesn’t make a habit of ending responses with unnecessary follow-up questions or overenthusiastic flourishes anymore. Everything feels cleaner, shorter, and less overformatted.

While GPT-5.5 Instant is available to all users, enhanced personalization from past chats, files, and connected Gmail accounts is currently rolling out only to Plus and Pro users on the web. OpenAI says mobile support is coming soon, with plans to expand the feature to Free, Go, Business, and Enterprise users in the coming weeks.

GPT-5.5 Instant is not the kind of update that produces dramatic viral videos. Instead, it seems designed to remove friction from everyday use — making ChatGPT feel calmer, cleaner, and a little less eager to over-perform every second of the conversation. I prefer it.

That may not sound revolutionary, but for people who use AI tools every day, it could end up being one of OpenAI’s most important upgrades yet.

Google logo on a black background next to text reading 'Click to follow TechRadar'


Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.


Purple circle with the words Best business laptops in white

Graham is the Senior Editor for AI at TechRadar. With over 25 years of experience in both online and print journalism, Graham has worked for various market-leading tech brands including Computeractive, PC Pro, iMore, MacFormat, Mac|Life, Maximum PC, and more. He specializes in reporting on everything to do with AI and has appeared on BBC TV shows like BBC One Breakfast and on Radio 4 commenting on the latest trends in tech. Graham has an honors degree in Computer Science and spends his spare time podcasting and blogging.