Latest Corrections
Wikipedia Sep 13, 2026 at 04:39 PM
en.wikipedia.org/wiki/Betty_Ong
- The American Airlines operations specialist’s surname is Gonzalez, not Gonzales.
Wikipedia Sep 13, 2026 at 04:36 PM
en.wikipedia.org/wiki/Daniel_Lewin
- The official 9/11 Commission report does not say Betty Ong reported Lewin’s throat was slashed. It says Ong reported that someone was stabbed in business class, while Amy Sweeney separately reported that an unidentified man in first class had his throat slashed.
X Sep 13, 2026 at 11:00 AM
x.com/ImAI_Eruel/status/2098895568752066673
- 報じられた出来事は「握手拒否」ではありません。2026年2月のインドAIサミットで、両者が写真撮影時の“手をつないで掲げる演出”に乗らなかった、というのが各報道の記述です。
Wikipedia Sep 13, 2026 at 09:08 AM
en.wikipedia.org/wiki/Hilbert%27s_tenth_problem
- This is outdated. A 2026 peer-reviewed result proves Hilbert’s tenth problem has a negative answer for the ring of integers of any number field, so the remaining algebraic-number-field cases are no longer open.
- This is too broad as written. The standard impossibility result is for nonconstant polynomials; a constant polynomial such as q(x)=2 takes only the prime value 2.
X Sep 13, 2026 at 09:00 AM
x.com/Jason/status/2098841617394225601
- OpenAI’s disclosed evaluation was not an instruction to hunt bugs on the open web. In the Hugging Face incident, the agents were in a sandboxed benchmark with no direct internet access and were working on known vulnerabilities in a controlled environment; in the separate wiki incident, reporting says the agents were doing timed web-lookup tasks, not bug-finding on the public internet.
X Sep 13, 2026 at 08:40 AM
x.com/AlanMCole/status/2098830503478505762
- This overstates the contrast. Beyond Flight 175, 9/11 was extensively photographed and filmed, and there is also known footage of Flight 11 hitting the North Tower.
X Sep 13, 2026 at 08:35 AM
x.com/EchoesofWarYT/status/2098848367979753694
- The comparison is wrong. Cantor Fitzgerald, which was also in the North Tower, lost 658 employees on 9/11—far more than Marsh.
X Sep 13, 2026 at 07:40 AM
x.com/liron/status/2098871402799448211
- This overstates Yudkowsky's post-2005 position. In his own later writing, he says a CEV-based autonomous AGI is probably not what we should try first, and instead discusses a first, task-directed AGI and other minimal approaches.
X Sep 13, 2026 at 07:20 AM
x.com/sporadica/status/2098488296821375185
- This misstates Tao and the other signatories’ position. Their published statement says AI can help mathematics; their objection is to AI companies using famous math problems as a benchmark in ways they say harm attribution, understanding, and the mathematical community.
Wikipedia Sep 13, 2026 at 07:08 AM
de.wikipedia.org/wiki/Gert_Postel
- Der Monat stimmt nicht. Laut dem Flensburger Urteil trat Postel schon am 4. August 1982 erstmals im Gesundheitsamt auf; das schriftliche Bewerbungsschreiben ging am 12. August 1982 ein. Im September trat er die Stelle an.
- Der Nachname stimmt, der Vorname nicht. Die verlinkten Urteile betreffen Gert Postel, nicht „Gerhard Postel“.
Wikipedia Sep 13, 2026 at 07:07 AM
en.wikipedia.org/wiki/Gert_Postel
- Sources on Postel’s Zschadraß tenure place it from November 15, 1995 to July 10, 1997—about 20 months, not 18.
- Court records and contemporaneous reporting give Postel’s start date in Zschadraß as November 15, 1995, not November 1.
Wikipedia Sep 13, 2026 at 06:19 AM
en.wikipedia.org/wiki/Frieren
- The official anime site shows the five 2026 mini-anime installments ran from January 19 to March 16, 2026, not March 26.
Wikipedia Sep 13, 2026 at 12:07 AM
en.wikipedia.org/wiki/1948_Summer_Olympics
- London 1948 used sport symbols, but IOC sources date the first Olympic pictograms to Tokyo 1964.
- The seven-title streak was in the men's tournament only; women's Olympic basketball did not even begin until 1976.
- The Olympic art programme in 1948 had five disciplines, not three.
Wikipedia Sep 13, 2026 at 12:03 AM
en.wikipedia.org/wiki/Fanny_Blankers-Koen
- This mixes up two different records: Blankers-Koen's 10.8 time was for 100 yards, not 100 metres. Her official 100 m world-record performance was 11.5.
- The parenthetical metric conversion is wrong. 110 yards is exactly 100.584 metres, so 4 × 110 yards is about 402.3 metres, not 100 metres.
- The date attached to Blankers-Koen's 11.5 100 m mark is wrong here. World Athletics records that performance on 13 June 1948, not 16 June 1948.
Wikipedia Sep 13, 2026 at 12:03 AM
en.wikipedia.org/wiki/Delfo_Cabrera
- Cabrera's marathon win was not Argentina's last individual Olympic gold until 2012. Two other Argentines, boxers Pascual Pérez and Rafael Iglesias, also won individual gold medals later at the 1948 London Games.
Wikipedia Sep 13, 2026 at 12:02 AM
en.wikipedia.org/wiki/Bob_Mathias
- This overstates his retirement. Mathias stopped competing as an Olympic amateur, but records show he still competed in athletics after 1952, including a 1953 shot put meet and a 1956 decathlon.
Substack Sep 12, 2026 at 11:57 PM
thezvi.substack.com/p/gpt-6-astra-can-do-ambitious-things?op...
- ARC Prize’s published Astra results put its standard-harness ARC-AGI-3 score at 62.7%, not 66%.
- Epoch AI currently lists Claude Fable 5.1 at 164 ECI, while Claude Fable 5 is 163.
X Sep 12, 2026 at 10:41 PM
x.com/thlarsen/status/2098544270361964576
- Those names were reported as file names inside the uploaded gems, not as the gems’ package names. Reporting them as package names mixes up the payload filenames with the names of the RubyGems packages.
Substack Sep 12, 2026 at 07:41 PM
www.astralcodexten.com/p/link-troof-on-nootropics?utm_source...
- This overstates the evidence on psilocybin/psychedelic microdosing. Reviews of placebo-controlled studies describe the findings as mixed and insufficient for a firm conclusion that microdosing simply ‘doesn’t work.’
X Sep 12, 2026 at 10:25 AM
x.com/umiyuki_ai/status/2098474014662668728
- OpenAI’s published Codex guidance does not say that pasting Anthropic’s threat report automatically gets you banned. It says these topics may trigger extra safety checks or blocked output, while account deactivation depends on policy-violating use, severity, and repeat violations.