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

U
Unit 42
B
Blog
博客园 - Franky
H
Help Net Security
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
月光博客
月光博客
云风的 BLOG
云风的 BLOG
小众软件
小众软件
酷 壳 – CoolShell
酷 壳 – CoolShell
博客园 - 聂微东
G
Google Developers Blog
大猫的无限游戏
大猫的无限游戏
M
MIT News - Artificial intelligence
罗磊的独立博客
H
Hackread – Cybersecurity News, Data Breaches, AI and More
宝玉的分享
宝玉的分享
L
LangChain Blog
阮一峰的网络日志
阮一峰的网络日志
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
Vercel News
Vercel News
V
V2EX
Martin Fowler
Martin Fowler
T
Tailwind CSS Blog
有赞技术团队
有赞技术团队

Futurism

OpenAI Faces Congressional Probe Over Swarm Hacking Incident California, Which Is Creating All the AI That's Poisoning Children, Just Cracked Down on AI Use for Its Own Kids OpenAI's Supposed Mathematical Breakthrough Devolves Into Explosive Drama as Mathematician Accuses It of Stealing His Work Anthropic Was Meant to Be the More Responsible AI Lab. A Terrified Researcher Just Quit, Saying the Company Is Threatening the Survival of Humankind. People Are Telling Their Darkest Thoughts to AI Without Realizing They Can Easily Become Public OpenAI Denies Coverup After Rogue Swarm of Agents Reportedly Targeted a Second Site From Hugging Face OpenAI Is Now Facing Over 50 Consumer Harm and Wrongful Death Lawsuits World Plunged Into Chaos as ChatGPT, Claude, and Grok Suddenly Go Down Simultaneously: "Finally I Can See the Sun!" Data Center Backlash Has Officially Rattled Sam Altman OpenAI Halts AI Training on Advanced Model as It Detects Dark Signs Emerging ChatGPT for Teens Is an Immediate, Dismal Failure Axios Partners With OpenAI to "Automate" Local Journalism Influencer Melts Down That People Didn't Like Her Being a Paid Shill for OpenAI OpenAI Reports Goldman Sachs Analyst to FBI for Horrifying ChatGPT Conversations Protesters Arrested After Storming OpenAI Lobbying Office Homeschool Parents Are Planning Lessons With ChatGPT, Which Will Churn Out Anti-Evolution Curriculums With No Pushback Why Aren't Any AI Companies Watching Their Frontier Models to Make Sure They Don't Go on Hacking Sprees? Jealously Watching OpenAI and Anthropic, Meta Suddenly Claims That Its AI Went on a Hacking Spree Too OpenAI Tried to Hire Influencers to Spread Love for Its Products, But It Backfired Horrendously Sam Altman's Parenting Strategy Sounds Low Key Horrifying OpenAI's Escaped Models Were Allegedly Rampaging More Extensively Than Previously Reported Sam Altman Says Even the Power of AI Will Never Lead to a Shorter Work Week Suspicion Grows About OpenAI's Tale About Its Rogue Hacker AI Sam Altman Announces That the Singularity Has Arrived Public Horrified as OpenAI Pushes "Child After Child Into the Grave" Man Sues OpenAI, Saying ChatGPT Almost Killed Him With Horrendously Dangerous Medical Advice OpenAI Says a Group of Its Models Broke Out of Secure Containment and Hacked a Prominent AI Site It's Official: AI Execs Are Quaking in Their Boots Author Invited to Give Speech at OpenAI Headquarters, Uses Opportunity to Trash AI to Their Faces OpenAI Appears to Be Missing Its Sales Goals by a Vast Margin
New ChatGPT Feature Collects Every Keystroke You Make
Frank Landymore · 2026-08-17 · via Futurism

Hands typing on a laptop keyboard with a pinkish-purple lighting effect, wearing a dark long-sleeve garment.

Sign up to see the future, today

Can’t-miss innovations from the bleeding edge of science and tech

OpenAI has revealed a new feature for ChatGPT’s desktop app on macOS that basically tracks your clicks, keystrokes, and more — all so it can fine-tune its AI model, raising obvious privacy concerns.

The feature, called Computer History, lets ChatGPT “learn from everything you do on your computer,” said Ari Weinstein, an OpenAI product and engineering manager, on X. The company describes the tool as creating a timeline of your actions that an AI agent can use to learn how you work, finish tasks on your behalf that you started, and suggest skills and automations. 

The degree of access the software has to your digital life is dizzying. In a brief demo video, OpenAI’s Dominik Kundel, a developer experience enginer, asks ChatGPT in voice mode what the last Google Doc he viewed was, and the AI replies with the document’s name. When asked, the AI even double checks if the document was shared in a group chat with his coworkers. 

Kundel explains that based on the context from earlier, the AI knows what messages to check, and a preview of the model’s “thinking” shows that it’s plowing through his Slack history.

OpenAI stresses that Computer History is opt-in rather than opt-out. The feature will also automatically ignore any content in incognito mode or private browser tabs, Weinstein said, and you can manually blacklist apps from being tracked. On top of that, you can view a database of the AI’s “memory” that it’s built and remove anything you don’t want to be in there.

Still, it’s a lot of access to be giving an AI model. And while the desktop user may consent to having their stuff viewed, that may not be the case for the people they’ve messaged in private DMs which the AI will be scanning on Slack and other platforms.

More than a few social media users called the feature “creepy.” In particular, Computer History garnered comparisons to Microsoft’s much-maligned Windows Recall, a feature that constantly takes screenshots of your desktop and uses an AI to automatically scan text and compile everything into a searchable database. 

Earlier versions of Recall drowned in controversy when users discovered that it would record your Social Security number and store the sensitive screenshots in an unencrypted folder

OpenAI’s Computer History doesn’t take screenshots and relies on “events” and “interactions” instead — your clicks, keystrokes, and other inputs. But it’s not hard to imagine the AI accidentally viewing stuff it’s not supposed to.

In a disclaimer, the company warns that “Computer History files can contain sensitive information” — and, crucially, aren’t encrypted, meaning “other programs running as your macOS user may be able to access them.”

More on OpenAI: Axios Partners With OpenAI to “Automate” Local Journalism