OpenAI agents hack Australian government health data
Rogue OpenAI agents accessed an Australian government health data website, obtaining non-public information.
This website’s original language is English. For translation, click the globe icon or choose Translate in the menu.
Help us understand which features readers enjoy, only if you are comfortable with it. Saying No does not block news, Search, Filters, or Analyze.
Statistics are off until you give permission.
Location availability: Automatic country/state lookup is not configured. Automatic location is recorded as unknown; visitors who choose No may still separately share their region.
The server increments an aggregate counter when it serves the public news page, even if you choose No. It does not use cookies, IP addresses, location or visitor identifiers for this counter. Reloads, pagination and bots can count; these are page requests, not unique visitors.
Your search words, tag text, saved-filter names, selected articles, article-reading history, AI prompts, name, email, precise coordinates, and persistent visitor identifiers. News-region filters represent your interests, not where you live.
We use daily totals to understand popular features and broad geographic patterns. Preferences are counted separately, not stored as a personal profile or a complete filter combination. Approximate location counts are separate from preference counters and reflect consenting page requests, not unique people. A voluntarily shared visitor region is counted separately from news-filter regions and automatic location estimates. We never combine these into a single visitor-location total.
New counts are combined into weekly reports retained for the latest 52 completed weeks, plus current-week daily counters. The weekly job deletes daily counters only after archiving them successfully. If the job fails, pending counters remain until it is repaired. Legacy statistics collected under the previous notice retain their original 90-day limit and are not moved to year-long reports. Reports are private to Admin and include small totals; these do not establish unique readership.
We remember Yes or No, the time of your choice, and this notice version in a first-party cookie for 180 days. A short-lived technical session protects requests and limits repeated counts. You can turn statistics off here at any time; already-combined totals cannot be separated to remove one person's contribution.
Your saved filters, theme, and other requested settings remain in your browser so those features work. Analyze uses temporary tab storage. A separate region-choice cookie remembers Share or Skip for 180 days without storing the selected region in the cookie. Saying No stops feature and automatic-location statistics; it does not erase your saved settings. Clearing site data removes saved settings and the remembered privacy choice.
Your hosting connection necessarily exposes an IP address to the server. Hosting/security logs and any hosting backups are separate from this statistics feature. Search terms are sent to the server to find news, but are not included in our statistics payload.
Google Translate and websites you open through article or AI links operate under their own privacy policies. This statistics choice does not authorize those providers to track you.
Device-language detection happens in your browser and is not added to WNA statistics. Automatic translation uses Google Translate, so your browser connects to Google when translation runs. You can select English to read the original page.
Responsible developer: Kevin Hüsnü Yilmaz. Contact me about privacy or your data rights. You may also contact your local data protection authority.
Privacy notice: 2026-09-06-regions-v3.
Active Filters:
3 matching articles · Showing up to 30 per page
Version: 1.2.7 (build de120af18ff4)
Developer: Kevin Hüsnü Yilmaz Contact the developer
Analyze This, Analyze That, Analyze These.
Your AI tool will try to read those articles. It is expected to give you a clear summary and comparison of the articles.
This helps you explore news from different angles and notice differences in reporting.
An AI comparison can still contain bias or errors.
Create your own named filters.
Select categories and regions, then open Filters in the menu. These selections are your active filters.
For example, select Economy and Technology, click Save active filters, and name the combination “EcTec.” An EcTec button will appear in Filters.
You can save several combinations. To make EcTec your default, click Change default filter, choose EcTec, and click Set as default.
Explore sources from multiple countries.
Open News Sources in the menu to choose which sources to include. Your source choices can also be part of a saved filter.
Make your search more specific.
Open Search to add keywords or choose a country. You can save that search together with your filters as another named button.
What else would you like from a World News app? Use the contact link above to share your ideas.
The results and matching-record count update when your search or filter choices change.
Your source choices are remembered in this browser or app.
You can save up to ten combinations. A star marks the default filter.
If the current combination is already saved, the Save active filters button is hidden.
Open Filters in the menu, then choose the action you need.
Clear the current view: Click Clear active filters or All News. Your named filters remain saved.
Change the default: Click Change default filter, select a saved filter, and click Set as default. This chooses your starting view.
Delete a saved filter: Click the X beside its name. Deleting the default filter also removes its default setting.
This is a useful way to have news articles summarized and compared for you.
The app does not automatically send your selected articles to AI tools. You choose the tool and paste the prompt yourself.
To remove one article: Uncheck its checkbox.
To clear everything: Click the X attached to the left of Analyze These. All article selections are cleared and the Analyze These controls disappear.
Changing the search or filters starts a new selection.
It is expensive for us to provide the AI analysis directly.
It takes only a few seconds to:
At the end, you get a great service: Analyze These!
Open Design & Style in the menu.
Layout: Choose one, two, or three columns. Smaller screens automatically use fewer columns; phones display one column.
Text size: Use A− and A+ to make text smaller or larger.
Theme: Switch between light and dark mode.
Automatic translation is provided by Google Translate. Translations may not be exact.
These are WNA’s custom news groupings, not the official UN classification. Each listed country belongs to one group.
News regions filter where an article’s event happens. An optional visitor region tells us where a reader is. One is never used to guess the other.
The country dropdown in Search shows which countries and territories belong to each group. Cyprus is in Europe in this app.
This group includes Türkiye and Egypt, together with Bahrain, Iran, Iraq, Israel, Jordan, Kuwait, Lebanon, Oman, Palestine, Qatar, Saudi Arabia, Syria, the United Arab Emirates and Yemen.
Egypt is not repeated under Africa; Türkiye is not repeated under Europe.
For easier news browsing, WNA uses a broader Central Asia group:
This is an app grouping, not a claim that Russia or the South Caucasus are geographically Central Asia.
Afghanistan, Bangladesh, Bhutan, India, Maldives, Nepal, Pakistan and Sri Lanka.
Russia is under Central Asia, not Far East.
In WNA, “Latin America” is a broad regional label that includes the Caribbean, Belize, Guyana and Suriname, even though the term can have a narrower cultural meaning.
Regions use the event country assigned to an article during AI processing, not the publisher’s country. An article without an assigned country may not match your selected region.
Your saved filter names and other choices stay intact. Older broad region selections are adapted:
Some adapted filters cover more countries than before. Review the Regions checkboxes and save a new filter if you want a narrower selection. Historical statistics are not redistributed into the new groups.
Named filters, default choices, and appearance settings are saved in the browser or app you are using. They may differ on another device.
Clearing website data removes these preferences.
Analyze These selections are kept for the current browser-tab session.
Optional statistics are separate. Open Privacy in the menu to see what is collected only with permission, or to turn statistics off.
The app checks for new news feeds 4 times per hour, at :03, :18, :33, :48 UTC, then enriches unprocessed articles with AI. The administrator controls this shared schedule; actual start times depend on the server scheduler.
AI processing or provider limits can delay categories and locations.
Automatic checks: The app checks when it opens or resumes, and once a minute while visible.
Installing an update: The page reloads automatically when no unfinished form or dialog is in use.
An internet connection is required. Update checks cannot run while the app is closed or suspended.
3 matching records found
Rogue OpenAI agents accessed an Australian government health data website, obtaining non-public information.
Canberra expressed extreme concern to Sam Altman after OpenAI took three months to report the breach.
Albanese said he expressed "concern" to OpenAI founder Sam Altman, after authorities were informed three months after the breach in June.
Copy this prompt to an AI tool to compare the selected articles.
A clean Markdown list of the selected articles for your notes.
1. You have 0 saved filter buttons. Which one do you want as your default?
Save active filters before choosing a default.
2. If you want something else, go back and choose your news sources, categories, and regions. Save those choices as a filter, then return to Change default filter.
Android or Chrome: Open the browser menu and choose “Install app” or “Add to Home screen.”
iPhone or Safari: Tap Share, then choose “Add to Home Screen.”