Brazil's electoral court has fined Meta daily. HN's top reply said the headline's evidence was numerology
Meta AI, asked to pick a number from 1 to 25, answered 13 and 22, the ballot numbers of Lula and Flávio Bolsonaro. HN's top comment found that exhibit risible.
Brazil's electoral court ordered Meta to apply identical criteria to both campaigns, with a fine of R$500,000 a day. Watchdog groups filed new screenshots as evidence.
The thread split the exhibits into laughable ones and one that a commenter said was real.
What the petition says, and what it concedes
The Panamerican Dispatch, translating from Brasil 247, reports that Justice Marco Aurélio Bellizze of the Superior Electoral Court ordered Meta to apply identical criteria to Lula's and Flávio Bolsonaro's campaigns and to prove compliance. The full court upheld the order. A new 24-hour deadline followed on Sunday, with a fine of R$500,000 a day, rising to R$1 million.
Sleeping Giants Brasil and CTRL+Z filed new screenshots. A WhatsApp GIF search for "brasil" returned an image of Bolsonaro showing his ticket number. A search for "ladrão" (thief) returned a caricature of Lula in a prison uniform. On Instagram, "lula" autocompleted to "#lulaladrão" first. The two groups say they have not proven bias and want the court to examine ranking. Meta denies favoring anyone and blames technical failures.
The top reply: the weakest exhibit sank the whole file
"My confidence in the validity of their claims tanked right about here… can we choose less stupid hills to die on than numerology and GIFs? Headline is not remotely supported by the evidence." — a commenter who quoted the petition's number-picking paragraph
The part of the case that held up
Another commenter agreed that the example was weak, then argued the real finding sits elsewhere.
"When they are asked to produce campaign material, they spontaneously produce it with Bolsonaro's number (22), even when not asked. But then they refuse to produce it with other numbers like Lula's 13." — nextaccountic, who linked a Brazilian forum thread of examples
That claim is not in the petition as reported, and the thread offered no reproduction of it. What the same commenter said would settle it is leaking the system prompt of the WhatsApp assistant. Until then the strongest claim in the case is an unconfirmed report.
Get breakdowns like this every day
The top Hacker News stories, unpacked — by email, Telegram, or RSS.
JetBrains launched Thinkrail, an agent IDE, and HN spent its thread on the README
The open-source app runs several coding-agent sessions in separate git worktrees. The top comment refused to take it seriously because the README looked AI-written.