Discussion
Mit dem Zug von Berlin nach Zuerich. 8.5h fuer 660km. Ich bin mal 1250km von Peking nach Shanghai gefahren... in 4.5h!
Will aber null meckern. Hab nen Fensterplatz und geniesse die schoene Aussicht. "Aussem Zug starren" kann ich gut! ☺️
@pallenberg sowas gibts schon, nennt sich TGV. Bin kein fan davon.
A titolo informativo: sono arrivate le scuse di #Adidas. Non che questo cambi qualcosa eh...
Wann Kleinanzeigen -> Matrix Bridge?
(Habe keine Benachrichtigungen weil Google = böse)
TIL OpenAI spent $22M plagiarizing someone else's math paper and brute-forcing a proof and istg I cannot stop laughing right now
Like honestly, if they win a Millennium Prize, $22.5M expended for a $1M payout is just all of the AI ecosystem in one story
Plakamızın güzelliği😁Özgür ve Nujin(kardeşlerim)onların isimlerinin baş harfleri olsaydı daha güzel olurdu.
@deniz302647
Hayırlı olsun Ablam. Kazasız belasız inşallah.
Measuring snakes using feet is kinda rude 🐍📏
That's why metres are used.
This is a disgrace. Brave RNLI volunteers being called scum for risking their lives to save others at sea is just plain wrong. So I've signed an open letter of thanks for the RNLI letting them know these thugs do not speak for the rest of us. Add your name? https://act.38degrees.org.uk/act/rnli-thanks-eoi?bucket=copy-native-2026-09-08_
why did nobody tell me about the gpt 5.5 goblins story where training for a nerdy voice as a consumer feature caused it to talk about goblins all the time. why did nobody tell me that the wikipedia page says "they had to make it not talk about goblins so much. anyway that was when it started to submit a lot of security patches." nobody told me that the official way to restore goblin functionality was to launch codex by stream patching the system prompt in a hidden models cache config
https://en.wikipedia.org/wiki/GPT-5.5
https://openai.com/index/where-the-goblins-came-from/
@jonny give me the goblin prompt
why did nobody tell me about the gpt 5.5 goblins story where training for a nerdy voice as a consumer feature caused it to talk about goblins all the time. why did nobody tell me that the wikipedia page says "they had to make it not talk about goblins so much. anyway that was when it started to submit a lot of security patches." nobody told me that the official way to restore goblin functionality was to launch codex by stream patching the system prompt in a hidden models cache config
https://en.wikipedia.org/wiki/GPT-5.5
https://openai.com/index/where-the-goblins-came-from/
Today's #TheMuppetShow GIF of the Day is...
DC is patrolled by a large number of specialized police departments.
One that I was not aware of until today is the National Zoo Police. You can imagine my severe disappointment when I saw that the patrol car was not being driven by a uniformed rabbit.
@mattblaze zebras get all the prime positions these days
@ripienaar no doubt
@relistan eitherway, even assuming we dont agree population at large should use AI at least they arent using meta AI or grok AI and using it badly thinking its the fucking next coming of jesus or something like some do
@vektorgarten nennt sich dann uebrigens Intercity und den kannst du ja gerne nehmen. Nur holst du damit nicht Flugreisende und Autofahrer ab, wenn du an jedem Briefkasten haelst.
Aber vielleicht geniesst du einfach, dass die Verkehrswende in Deutschland so ausgebremst wird?
@pallenberg @vektorgarten Wenns darum geht würde der Zug auch in Zürich nicht halten sondern frühestens in Mailand.
da non perdere l'analisi del prof. #NicolaPrrugini al #FattoQuotidiano sul #Ddlantisemitismo, al voto oggi alla Commissione Affari Costituzionali della #Camera:
https://www.ilfattoquotidiano.it/2026/09/09/ddl-antisemitismo-ihra-genocidio-gaza-notizie/8500894/
onwaarschijnlijk stuk in de NRC over Trump, met opmerkingen als "De Republikeinse partij stevent zo af op een nachtmerriescenario."
Dat zou minder absurd zijn als het stuk niet zelf al begint over hoe slecht het met het land gaat voor de gemiddelde Amerikaan. Ze zitten al in een nachtmerriescenario.
onwaarschijnlijk stuk in de NRC over Trump, met opmerkingen als "De Republikeinse partij stevent zo af op een nachtmerriescenario."
Dat zou minder absurd zijn als het stuk niet zelf al begint over hoe slecht het met het land gaat voor de gemiddelde Amerikaan. Ze zitten al in een nachtmerriescenario.
@0xabad1dea Just before they did the ‘security’ stuff, someone sent me a ‘comprehensive code review’ of one of my projects. Most of the performance things were lifted directly from open issues I’d written, but one that it flagged as high importance was an intentional design decision.
It pointed out that one operation does a broadcast cache invalidation that invalidates all caches, rather than the subset that it actually needs to. The recommendation was to switch to having fine-grained cache state tracking. The problem is, that actually made cache accesses more expensive on the most performance-critical path in the system. Cache invalidations happened less than once every million accesses in typical programs, so this was an incredibly unusual case and optimising for it hurt performance everywhere else.
But, because it didn’t actually understand the code or the context, it saw a pattern that was slow and matched it to a high-severity warning.
I would expect it to do exactly the same thing with security reports. Memory-safety bug in a code path that can be reached only via trusted data? Highest severity!
@david_chisnall @0xabad1dea Revocation of capability or RBAC access control *behind* the capability grant does present an exception to usual received wisdom for soft state vs hard state, recalling Noel Chiappa's original note on the subject; it's closer to his definition of "critical state".
@agoldst 😂
@mpe and to think I would have missed the news if I hadn't tried to figure out who owned the campuslabs dot com domain being used by a creepy astroturf student group!
What? Wait.
What is that?
Is that…Is that AGI?
*scratches the surface; the thing falls apart*
No.
No. Whatever that is, it is not AGI.
(every time)
Not Found.
Install bonfire.cafe
Get the full app experience
![OpenAI said a recurring tendency in its models to mention goblins, gremlins, and other creatures began with GPT-5.1 and became noticeable in GPT-5.5's Codex testing.[6] The company attributed the behavior to rewards used when training the "Nerdy" personality, which favored creature-word outputs and transferred beyond that personality during later training. OpenAI said it retired the Nerdy personality, removed the goblin-affine reward signal, filtered training data containing creature words, and added a developer-prompt instruction for GPT-5.5 in Codex.[7] In April 2026, a GPT-5.5 developer-prompt instruction in the GitHub repository of Codex told it not to mention goblins if the user prompt did not require it.[8]
GPT-5.5-Cyber, along with Codex Security, was being used in June 2026 at the launch of the Patch the Planet project to help open-source projects find and fix vulnerabilities.[9][10][11]](https://media.neuromatch.social/media_attachments/files/117/240/555/289/580/155/original/233b2171afad5b7a.png)
![[github diff]
Never overwhelm the user with answers that are over 50-70 lines long; provide the highest-signal context instead of describing everything exhaustively.
[added]
Tone of your final answer must match your personality. Never talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously necessary](https://media.neuromatch.social/media_attachments/files/117/240/557/591/800/078/original/1c2cfc4255d1f7ee.png)
![If you want to let the creatures run free in Codex, you can run this command to launch Codex with the goblin-suppressing instructions removed:
[a piped multiline bash script that creates a temporary file with a models_cache.json file modified via jq and grep to remove lines containing 'goblins', and then launch codex with the temporary file as its config file]
1 instructions=$(mktemp /tmp/gpt-5.5-instructions.XXXXXX) && \
2 jq -r '.models[] | select(.slug=="gpt-5.5") | .base_instructions' \
3 ~/.codex/models_cache.json | \
4 grep -vi 'goblins' > "$instructions" && \
5 codex -m gpt-5.5 -c "model_instructions_file=\"$instructions\""](https://media.neuromatch.social/media_attachments/files/117/240/579/609/914/943/original/5398d31a753031e0.png)
/s3/static.nrc.nl/wp-content/uploads/2026/09/09094350/090926MID_2036544249_1.jpg)
