TRENDING
A person sweeping a metal detector across a wet sandy beach with a sea fort on the horizon
November 22, 2026
How to Write YARA Rules With YARA-X to Hunt Malicious Python .pth Startup Files and Measure False Positives Before You Trust Them
A peeling black-and-white metal door standing ajar, with a dark empty hall and bright windows beyond
November 22, 2026
AWS’s “Found a Zero-Day on Their Own” Post Never Names the Product; the Advisory Behind Its Link Shows a Server That Ships With No Login, a README Now Saying Keep It Off the Internet, and a Fix Merged October 19 That PyPI and Docker Hub Still Do Not Carry
A hand paging through index cards in an open wooden library card-catalog drawer labeled Agriculture, with a second drawer open below to the right
November 22, 2026
Microsoft Tells Customers Unauthorized Parties Read Admin Email Addresses and Tenant Metadata Through an Entra Connect Health Flaw Introduced July 23, but Gives No Tenant Count or Start Date, and Its Own Docs Send Tenant Security Notices to the Same Kinds of Contacts
Swimmers in separate lanes of a blue swimming pool, seen from above the water, with lane lines running across the picture
November 22, 2026
How to Size a PostgreSQL Connection Pool in Python: Find Where Throughput Stops Growing, Split max_connections Across Workers, and Stop Slow Handlers From Hoarding Connections
Workers in caps lean over a 1913 Ford assembly line, each assembling parts as they pass on the moving line
November 22, 2026
OpenAI’s Performance Factory Turns Regression Response Into a Task Queue for Agents, but Its Numbers Stop at 600 Root-Caused Cases in 1,700 Investigations and a 70% Grade on 100 Suggestions, and in a Toy World Each of Its Reviewer’s Two Counterfactual Questions Can Return a Confident Wrong Answer
22 Nov 2026
SXZ.io SXZ.io
  • Home
Search the Site
Popular Searches:
Technology Amazon AI
Recent Posts
The sun setting over the sea behind a pyramid-shaped stone monument at Cape Soya, the northernmost point of Japan
How to Retire an API Endpoint in FastAPI: Send Deprecation and Sunset Headers, Find the Clients Still Calling It, Schedule Brownouts, and Block an Early Removal With oasdiff
November 22, 2026
Brass sextant in an open wooden case inside a museum display, an instrument for measuring the angle to the horizon
MIT Technology Review’s “Most Misunderstood Graph” Update Reports METR’s Judgment Findings Mostly Fairly, but It Never Says Each Point Is a 50% Success Rate or Gives the New 3.6-Month Doubling Time and 2.1 to 4.9 Month Range, and It Credits OpenAI With a Comparison METR Built
November 22, 2026
Brass telescope on a wooden tripod inside a white observatory dome with the roof slit open to the sky
Rackspace Says Leaked Credentials Let an Intruder Act as Legitimate Users in Some Spot Customer Organizations and Reach Their Kubernetes Secrets, a Platform9 Co-Founder Names a Breached Observability System as the Root Cause, and the Status Page Says Only “System Maintenance Operations”
November 22, 2026
SXZ.io SXZ.io
  • Home

Categories

Articles 310 Posts
News 316 Posts
Learning Hub 282 Posts
Home/Articles/Google Splits Gemini Chat History From Training, but “History Off” and Temporary Chats Now Keep Chats Around Two Months Instead of 72 Hours, and the New Training Switch Leaves Chats Open to Review for Other Purposes
Articles

Google Splits Gemini Chat History From Training, but “History Off” and Temporary Chats Now Keep Chats Around Two Months Instead of 72 Hours, and the New Training Switch Leaves Chats Open to Review for Other Purposes

Google’s help pages show Gemini’s history-off and temporary-chat retention moving from 72 hours to around two months within hours of a blog post that says the rollout starts in the coming weeks, and...

November 18, 2026 15 Min Read
40

On November 16 Google announced new controls for the Gemini app. One lets people keep their chat history while opting out of model training. The other is a safety change that the announcement states in a single sentence: “Now, if you turn your new History setting off, future chats will be saved for around two months.” The Register spotted what that sentence leaves out. Before the change the figure was 72 hours, which makes the new window about 20 times longer.

Table Of Content

  • What the two copies of the hub say
  • When the text changed
  • What the announcement leaves out
  • Temporary chats moved with History-off chats
  • The stated reason stayed the same while the window grew 20-fold
  • The training switch is narrower than the old history switch
  • The page changed before the rollout the blog describes
  • How the new window compares
  • Google’s other “two months”
  • What it means if you use Gemini
  • Picking a setting
  • History on, training off
  • History off or a temporary chat
  • Feedback
  • Work and school accounts
  • If you are in the EU or the UK
  • Reproduce the check
  • Download two archived copies
  • Print the retention and review sentences
  • Find when the text changed
  • Check the 20-fold arithmetic
  • Questions Google has not answered
  • Sources

I checked the claim against Google’s own pages instead of the coverage. The Gemini Apps Privacy Hub was archived on November 15, before the announcement, and it is live today, so the two copies can be compared sentence by sentence, and a few more archived copies show when the text changed. The comparison confirms The Register’s figure and adds findings that neither the blog post nor the news story mentions: temporary chats moved by the same factor, the new training switch is narrower than the old history switch, and the help page changed within hours of the blog post even though the post says the rollout starts “in the coming weeks.”

I did not test the Gemini app, and Google did not answer The Register’s questions, so everything below is about what Google’s documents say, not about what the product does. The method is the one we used on OpenAI’s usage policy rewrite and on Google’s privacy-policy change over Gmail, Drive and Photos ads: read the page itself, diff it against an archived copy, and quote both.

What the two copies of the hub say

Table 1 sets the old and new wording side by side. The old column is the archived copy of November 15 at 09:46 UTC, which Google had last updated on October 26. The new column is the live page on November 18, which Google had last updated on November 17. Every entry in the two right-hand columns is a quotation from the page.

Topic Old copy (November 15) Live page (November 18)
Name of the switch “Keep Activity” “History”
Chats kept when the switch is off “72 hours” “around two months”
Temporary chats (the same sentence as the row above) “72 hours” “around two months”
Gemini Live chats when the switch is off “up to 72 hours so Google can provide the service and process any feedback” “around two months so Google can provide the service, process any feedback, and protect Google, its users, and the public”
Example reason for keeping the chats “so it is available in case of a potential system failure” “so they are available in case of a potential system failure”
Context used to answer you “using the last 24 hours of your chat as context” “using the last 24 hours of your chat as context”
How to stop review “To stop your future chats from being reviewed to improve Google services, turn off your Keep Activity setting.” “You can stop your future chats from being reviewed to train generative AI models anytime in your Gemini History settings.”
Temporary chats and human review “not used to improve Google AI with help from human reviewers” “not used to train generative AI models with help from human reviewers”
What the settings page controls “control whether your data is used to improve Google AI” “control whether your data is used to train generative AI models”
Reviewed chats “retained for up to three years” “retained for up to three years”

Two rows matter before the detail. The second and third rows show that temporary chats are covered by the same sentence as chats saved while the switch is off. The three rows about review and the settings page show the vocabulary moving from “improve” to “train,” which the section on the training switch below takes up.

When the text changed

The Wayback Machine lists 19 successful captures of the hub from November 14 to 17. I fetched eight of them, from November 15 to 17, and they split cleanly: five carry the old wording and three carry the new one. Table 2 gives the timeline.

Time (UTC) Event Evidence
Nov 15, 09:46:35 Old wording; page last updated Oct 26 archived capture
Nov 16, 17:00:00 Blog post published the post’s own markup
Nov 16, 19:48:32 Latest capture with the old wording (“72 hours” four times) archived capture
Nov 16, 20:48:53 Blog post last modified the post’s own markup
Nov 16, 23:22:02 First capture with the new wording (“around two months” five times) archived capture
Nov 17, after 10:15 Hub edited again; page last updated moves to Nov 17 capture of 10:15 against the live page
Nov 17, 17:00 The Register publishes its story the story’s byline

The edit therefore landed inside a window of 3 hours 33 minutes 30 seconds on the day of the blog post. The blog’s last-modified time sits inside that window, and I draw no conclusion from that beyond the order of events. The second edit, on November 17, renamed leftover “Activity” wording to “History,” added an answer about texting Gemini for tasks, and rewrote a remote-browser paragraph. The two-month retention sentences did not change in it.

What the announcement leaves out

Temporary chats moved with History-off chats

The blog post and the news coverage describe the change as applying when History is off. The help page covers two cases with one sentence. The old copy says “Temporary chats and chats you have when Keep Activity is off are retained with your account for 72 hours.” The live page says “Temporary chats and chats you have when your History setting is off are retained with your account for around two months.” A temporary chat is the option a person reaches for when one conversation should leave nothing in their history, and its documented retention moved by the same factor.

The temporary-chat how-to page still says a temporary chat “won’t appear in your Gemini Apps Activity,” which is the old name, so the rewrite is not finished. I found no sentence on the hub, on that page or on the page about managing Gemini History that describes a way to shorten the window.

The stated reason stayed the same while the window grew 20-fold

The sentence that explains why Google keeps these chats is the old one with the number swapped. Old: “Google stores your chats with your account for 72 hours so it is available in case of a potential system failure.” New: “Google stores your chats with your account for around two months so they are available in case of a potential system failure.” The page also says Gemini responds to you “using the last 24 hours of your chat as context.”

So answering needs one day of context, the example reason is a system failure, and the stored window went from three days to about two months. The last step of the reproduction below puts numbers on it: 24 hours was a third of the old window and is under 2 percent of the new one. If the added weeks serve the purpose the blog post names, to “protect Google, its users, and the public,” the page does not say what that purpose needs that 72 hours did not provide.

The training switch is narrower than the old history switch

The headline feature is that people can now keep history and opt out of training. The old page tied review to the single Keep Activity switch: “To stop your future chats from being reviewed to improve Google services, turn off your Keep Activity setting.” The new page words the matching sentence more narrowly: “You can stop your future chats from being reviewed to train generative AI models anytime in your Gemini History settings.” For the new Chat content setting it then says what stays on when training is off: “Your future chats won’t be used to train generative AI models, unless you provide feedback. Future chats will be reviewed and used to improve Google services in other ways, and to protect Google, its users, and the public. Chats reviewed by service providers are separated from your account and kept for up to 3 years.”

That is a clear statement, but it is not the one in the blog post, which says the new settings “let you keep your Gemini chat history, while opting out of generative AI model training.” Two sentences set the outer limit. One has been on the page all along: “Your Gemini settings don’t control processing of your chats to create anonymized data to improve Google services.” The other was widened. The old copy said “Even if your Keep Activity setting is off or you use temporary chats, Google still uses your chats to respond to you and help protect Google, our users, and the public, including with help from human reviewers,” and the live page restates it for every setting: “Regardless of your settings, Google uses your chats to respond to you and help protect Google, our users, and the public, including with help from human reviewers.”

Human review of these chats was therefore already disclosed before the change. What is new is that the History-off sentence itself now pairs the two-month window with human reviewers. The page says the review is partial: “Human reviewers (including trained reviewers from our service providers) review some of the data we collect for these purposes.” And reviewed chats outlive any deletion: “Reviewed chats are saved for up to 3 years, even if you delete them.”

The page changed before the rollout the blog describes

The blog post says “We’ll start rolling out these changes globally in the Gemini app in the coming weeks.” The help page that states the new retention as the rule was live that same evening, and the page about managing Gemini History uses the new wording too. For a person who has not yet received the new settings, the pages do not say whether 72 hours or two months applies. The Register reports that Google did not respond to its questions, and I did not contact Google.

How the new window compares

Two other assistants document a floor for chats that are not saved to history. I read both pages on November 18.

Assistant and mode What the page says
Gemini, History off or a temporary chat (copy of November 15) retained with your account for “72 hours”
Gemini, History off or a temporary chat (live page) retained with your account for “around two months”
ChatGPT, temporary chat “OpenAI may keep a copy of a temporary chat for up to 30 days for safety purposes.”
Claude, incognito chat “they are retained for 30 days for safety, or longer in accordance with your organization’s data retention policy”

The comparison is loose. The pages use different words, and the Claude page adds that enterprise customers can set a longer period. Even so, the direction is plain: before November 16 Gemini’s documented floor was about a tenth of both rivals’ 30 days, and it now sits at about twice theirs. Our earlier post on Claude’s unified memory covers the personalization side of the same question, which is what an assistant keeps and uses beyond a single chat.

Google’s other “two months”

Google’s general retention page uses the same phrase for something else: after data is deleted, “This process generally takes around 2 months from the time of deletion. This often includes up to a month-long recovery period in case the data was removed unintentionally.” The same page says of backup storage: “Data can remain on these systems for up to 6 months.” The Gemini page does not say whether the History-off window is a retention period that starts a deletion process afterward or already includes it, and the two readings would differ by months.

What it means if you use Gemini

Picking a setting

These notes use only what the pages say.

History on, training off

The new Chat content setting is the control. With it off, the page says chats are not used for training unless you send feedback, and are still reviewed and used to improve Google services in other ways. I found no statement of its default on the pages I read. The page about managing history does say “If you’re 18 or over, History is on by default.”

History off or a temporary chat

Both now mean around two months with your account, for the purposes the page lists, with no documented way to shorten it. The page’s own advice applies: “Please don’t enter confidential information that you wouldn’t want a reviewer to see or Google to use to improve our services, including machine-learning technologies.”

Feedback

Sending feedback changes the picture for that conversation. The page says “Reviewed feedback, associated conversations, and related data are retained for up to 3 years, disconnected from your Google Account.” The old copy has the same sentence, so this part did not change.

Work and school accounts

The page about managing history says “This article covers Gemini history for a personal account. If you use Gemini Apps with a work or school account, your Gemini History settings are controlled by your Google Workspace administrator.” I did not look at what Google says about retention for those accounts.

If you are in the EU or the UK

The same hub lists Google’s legal bases for processing Gemini chats under EU and UK law. For processing chats to maintain the safety and reliability of the service it relies on legitimate interests, with examples that include “fixing bugs and troubleshooting failures” and “updating safety classifiers and model filters.” The GDPR’s storage-limitation principle says personal data must be “kept in a form which permits identification of data subjects for no longer than is necessary for the purposes for which the personal data are processed,” with exceptions for archiving, research and statistical purposes. Whether two months is necessary for those purposes is a question for Google and for regulators, and I make no finding on it. The page tells readers they may have the right to object and says “you can create a request in our Help Center.”

Reproduce the check

You need Python 3 and curl. The archived copies do not change, so the output below is repeatable. The live page can change, and it was edited twice in two days. The archive rate-limits bursts of requests, so if a download returns a short error page instead of the hub, wait a minute and run it again.

Download two archived copies

These commands save the hub as archived before the announcement and the evening of the announcement. Each file is about 1.7 MB.

curl -sL -A "Mozilla/5.0" -o old.html "https://web.archive.org/web/20261115094635/https://support.google.com/gemini/answer/13594961"
curl -sL -A "Mozilla/5.0" -o new.html "https://web.archive.org/web/20261116232202/https://support.google.com/gemini/answer/13594961"
ls -l old.html new.html

Print the retention and review sentences

The script strips the markup, keeps one text line per block, and prints every line that mentions 72 hours or around two months, plus the sentence about stopping review. The counts show four old retention lines against five new ones.

# gemini_lines.py
"""Print the retention and review sentences of saved copies of the Gemini Apps Privacy Hub."""
import html
import re
import sys

sys.stdout.reconfigure(encoding="utf-8")

RETENTION = re.compile(r"72 hours|around (?:2|two) months", re.I)
REVIEW_SWITCH = re.compile(r"stop your future chats from being reviewed", re.I)


def page_lines(path):
    raw = open(path, "rb").read().decode("utf-8", "replace")
    raw = re.sub(r"(?is)<(script|style|noscript|svg).*?</\1>", " ", raw)
    raw = re.sub(r"(?is)<br\s*/?>|</p>|</h\d>|</li>|</div>|</tr>", "\n", raw)
    text = html.unescape(re.sub(r"<[^>]+>", " ", raw))
    lines = (re.sub(r"\s+", " ", line).strip() for line in text.split("\n"))
    return [line for line in lines if line]


for path in sys.argv[1:]:
    lines = page_lines(path)
    updated = next((line for line in lines if line.startswith("Last updated")), "no date found")
    kept = [line for line in lines if RETENTION.search(line)]
    old = sum("72 hours" in line for line in kept)
    print(f"== {path} | {updated}")
    print(f"retention sentences (72 hours: {old}, around two months: {len(kept) - old})")
    for line in kept:
        print("  -", line)
    review = [line for line in lines if REVIEW_SWITCH.search(line)]
    print(f"review-switch sentences ({len(review)})")
    for line in review:
        print("  -", line)

Run it with python gemini_lines.py old.html new.html:

== old.html | Last updated: October 26, 2026
retention sentences (72 hours: 4, around two months: 0)
  - Temporary chats and chats you have when Keep Activity is off are retained with your account for 72 hours and used to:
  - protect Google, our users, and the public. For example, Google stores your chats with your account for 72 hours so it is available in case of a potential system failure.
  - Your future chats won’t appear in your Activity, and won’t be used to train our AI models, unless you choose to send Google feedback. Future chats are still saved for 72 hours so Gemini can respond to you, process your feedback, and protect Google, its users, and the public.
  - If Keep Activity is off, your Gemini Live chats are stored in your Google Account for up to 72 hours so Google can provide the service and process any feedback. Learn more .
review-switch sentences (1)
  - To stop your future chats from being reviewed to improve Google services, turn off your Keep Activity setting. If that setting is on, don’t enter data that’s confidential or that you wouldn’t want a reviewer to see or Google to use to improve its services. You can also use temporary chats , which are not used to improve Google AI with help from human reviewers. Learn more about temporary chats .
== new.html | Last updated: November 16, 2026
retention sentences (72 hours: 0, around two months: 5)
  - Even if Gemini History is turned off, future chats will still be saved for around 2 months and used to respond to you, process your feedback, and protect Google, its users, and the public, with help from human reviewers.
  - Temporary chats and chats you have when your History setting is off are retained with your account for around two months and used to:
  - help protect Google, our users, and the public. For example, Google stores your chats with your account for around two months so they are available in case of a potential system failure.
  - Future chats will still be saved for around 2 months and used to respond to you, process your feedback, and protect Google, its users, and the public, with help from human reviewers.
  - If History is off, your Gemini Live chats are stored in your Google Account for around two months so Google can provide the service, process any feedback, and protect Google, its users, and the public. Learn more .
review-switch sentences (1)
  - You can stop your future chats from being reviewed to train generative AI models anytime in your Gemini History settings. Learn more about how to manage your settings. You can also use temporary chats , which are not used to train generative AI models with help from human reviewers. Learn more about temporary chats .

Find when the text changed

The archive’s index lists the captures around the edit. This request returns the three captures between 19:00 and midnight UTC on November 16:

curl -s "https://web.archive.org/cdx/search/cdx?url=support.google.com/gemini/answer/13594961&from=20261116190000&to=20261116235959&output=txt&fl=timestamp,statuscode&filter=statuscode:200"
20261116194832 200
20261116232202 200
20261116233335 200

This script fetches captures by timestamp and classifies each by its wording. If a timestamp prints “could not be fetched (rate limited?),” wait a few minutes and run the script again for that timestamp alone. Run it with python capture_scan.py 20261115094635 20261116194832 20261116232202 20261117101533:

# capture_scan.py
"""Check Wayback Machine captures of the Gemini Apps Privacy Hub for the old and the new retention wording."""
import html
import re
import sys
import time
import urllib.request

PAGE = "https://support.google.com/gemini/answer/13594961"


def fetch(stamp):
    url = f"https://web.archive.org/web/{stamp}/{PAGE}"
    for _ in range(6):
        request = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0"})
        try:
            with urllib.request.urlopen(request, timeout=60) as response:
                return response.read().decode("utf-8", "replace")
        except Exception:
            time.sleep(25)
    return None


for stamp in sys.argv[1:]:
    raw = fetch(stamp)
    if raw is None:
        print(stamp, "could not be fetched (rate limited?)")
        continue
    raw = re.sub(r"(?is)<(script|style|noscript|svg).*?</\1>", " ", raw)
    text = re.sub(r"\s+", " ", html.unescape(re.sub(r"<[^>]+>", " ", raw)))
    old = len(re.findall(r"72 hours", text))
    new = len(re.findall(r"around (?:2|two) months", text))
    updated = re.findall(r"Last updated: ([A-Za-z]+ \d+, \d{4})", text)[:1]
    verdict = "old wording" if old and not new else "new wording" if new and not old else "mixed"
    print(f"{stamp}  72 hours: {old}  around two months: {new}  last updated: {updated[0]}  -> {verdict}")
    time.sleep(8)
20261115094635  72 hours: 4  around two months: 0  last updated: October 26, 2026  -> old wording
20261116194832  72 hours: 4  around two months: 0  last updated: October 26, 2026  -> old wording
20261116232202  72 hours: 0  around two months: 5  last updated: November 16, 2026  -> new wording
20261117101533  72 hours: 0  around two months: 5  last updated: November 16, 2026  -> new wording

The first two captures carry the old wording and the last two the new one, which brackets the edit between 19:48:32 and 23:22:02 UTC on November 16.

Check the 20-fold arithmetic

Google says “around two months” and does not give a number of days, so the script tries 59 to 62 days:

# window_math.py
"""How much longer is "around two months" than 72 hours, and how much of it is the 24 hours of context?"""
OLD_HOURS = 72
CONTEXT_HOURS = 24

print(f"old window: {OLD_HOURS} h; the {CONTEXT_HOURS} h of context is {CONTEXT_HOURS / OLD_HOURS:.1%} of it")
for days in (59, 60, 61, 62):
    hours = days * 24
    print(f"{days} days = {hours:>4} h: {hours / OLD_HOURS:4.1f} times the old window; "
          f"the {CONTEXT_HOURS} h of context is {CONTEXT_HOURS / hours:.1%} of it")
old window: 72 h; the 24 h of context is 33.3% of it
59 days = 1416 h: 19.7 times the old window; the 24 h of context is 1.7% of it
60 days = 1440 h: 20.0 times the old window; the 24 h of context is 1.7% of it
61 days = 1464 h: 20.3 times the old window; the 24 h of context is 1.6% of it
62 days = 1488 h: 20.7 times the old window; the 24 h of context is 1.6% of it

At 60 days the new window is exactly 20 times the old one, which matches The Register’s “20x,” and the 24 hours of context shrinks from a third of the window to under 2 percent of it.

Questions Google has not answered

  • What number of days does “around two months” stand for, and how does it relate to the roughly two months its general retention page gives for completing a deletion?
  • Which users have the new settings now, and which window applies to everyone else until they do?
  • What does the “protect” purpose need from a stored chat after the first 24 hours that it did not need before November 16?
  • Was the move of temporary chats to the same window intended, given that the announcement mentions only the History setting?
  • How many chats stored under History off or in temporary chats are reviewed by people during the window, and by Google staff or by service providers?
  • What is the default for the new Chat content setting, and does it differ by age or region?

Sources

  • Google, We’re updating controls in the Gemini app (November 16, 2026).
  • Google, Gemini Apps Privacy Hub: archived copy of November 15, archived copy of November 16 at 23:22 UTC and the live page.
  • Google, Manage and delete your Gemini history and Use temporary chats.
  • Google, How Google retains data we collect.
  • The Register, Opting out of Gemini tracking? Google will still hold your data 20x longer (November 17, 2026).
  • OpenAI, Temporary Chat FAQ. Anthropic, Using incognito chats.
  • General Data Protection Regulation, Article 5.

Tags:

Data PrivacyData RetentionGeminiGoogleTemporary Chats

Share

Upturned orange coffee cup on a saucer with roasted coffee beans spilling across a dark slate surface
Previous Post

Java4Shell Lets Malformed Base64 Overwrite the Heap in OpenJDK on AVX-512 Servers, but Vendors Rate It From 8.1 to 9.8, the CVE Text Names Neither Base64 Nor the CPU, and Eclipse Temurin and Ubuntu Had No Fixed Build Hours After Oracle’s Patch

Six painted Russian nesting dolls of different sizes standing in a row on a white background
Next Post

How to Find and Stop HTTP Request Smuggling: Smuggle a Request Past a Toy Proxy in Python, Test uvicorn and nginx, and Write a Strict Parser That Refuses Ambiguous Messages

No Comment! Be the first one.

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Latest
22 Nov
How to Retire an API Endpoint in FastAPI: Send Deprecation and Sunset Headers, Find the Clients Still Calling It, Schedule Brownouts, and Block an Early Removal With oasdiff
22 Nov
MIT Technology Review’s “Most Misunderstood Graph” Update Reports METR’s Judgment Findings Mostly Fairly, but It Never Says Each Point Is a 50% Success Rate or Gives the New 3.6-Month Doubling Time and 2.1 to 4.9 Month Range, and It Credits OpenAI With a Comparison METR Built
Trending
November 22, 2026
How to Retire an API Endpoint in FastAPI: Send Deprecation and Sunset Headers, Find the Clients Still Calling It, Schedule Brownouts, and Block an Early Removal With oasdiff
November 22, 2026
MIT Technology Review’s “Most Misunderstood Graph” Update Reports METR’s Judgment Findings Mostly Fairly, but It Never Says Each Point Is a 50% Success Rate or Gives the New 3.6-Month Doubling Time and 2.1 to 4.9 Month Range, and It Credits OpenAI With a Comparison METR Built
November 22, 2026
Rackspace Says Leaked Credentials Let an Intruder Act as Legitimate Users in Some Spot Customer Organizations and Reach Their Kubernetes Secrets, a Platform9 Co-Founder Names a Breached Observability System as the Root Cause, and the Status Page Says Only “System Maintenance Operations”
November 21, 2026
How to Write YARA Rules With YARA-X to Hunt Malicious Python .pth Startup Files and Measure False Positives Before You Trust Them
November 21, 2026
AWS’s “Found a Zero-Day on Their Own” Post Never Names the Product; the Advisory Behind Its Link Shows a Server That Ships With No Login, a README Now Saying Keep It Off the Internet, and a Fix Merged October 19 That PyPI and Docker Hub Still Do Not Carry
November 21, 2026
Microsoft Tells Customers Unauthorized Parties Read Admin Email Addresses and Tenant Metadata Through an Entra Connect Health Flaw Introduced July 23, but Gives No Tenant Count or Start Date, and Its Own Docs Send Tenant Security Notices to the Same Kinds of Contacts

Related Posts

Blue-lit server racks in a modern data center, illustrating the compute infrastructure behind the AI boom.
Articles

The AI Boom Is Spending Real Money Before Proving Real Returns

June 7, 2026
Technician working with a laptop beside server racks, representing enterprise AI retrieval infrastructure
Articles

Google’s Agentic RAG Push Makes Enterprise AI Less of a One-Shot Guess

June 7, 2026
A person with a laptop and smartphone, representing digital attention and AI-assisted work
Articles

AI Chatbots Are Making Attention a Design Problem

June 7, 2026
A customer-support representative wearing a headset against a dark studio background.
Articles

The Meta AI Support Hack Was a Plain Old Authorization Failure

June 7, 2026
SXZ.io SXZ.io
  • [email protected]

Categories

Articles
Learning Hub
News

All Rights Reserved by SXZ.io ©2026