Tootfinder

Widerstand gegen Online Safety Act in Großbritannien wächst
Mit dem Online Safety Act müssen auch Plattformen wie Reddit für eine Altersverifikation sorgen. Diese werden umgangen. Eine Petition hat viele Unterschriften.

from my link log —
UK Online Safety Act Notes for small sites.
https://russ.garrett.co.uk/2024/12/17/online-safety-act-guide/
saved 2024-12-17

Taylor Swift purchases her catalog of recordings, which were initially released through Big Machine Records, from Shamrock for an undisclosed nine-figure sum (Chris Willman/Variety)
https://variety.com/2025/music/news/taylor-swift…

Taylor Swift Shocker: Singer Buys Back Rights to First Six Albums
Taylor Swift has bought her catalog back from Shamrock and will reissue her original albums, while 'Reputation (Taylor's Version)' will still come out.

Flash floods in Pakistan kill 8 and 58 are rescued after deluge swept away dozens https://news.mongabay.com/short-article/2025/06/flash-floods-in-pakistan-kill-8-and-58-are-rescued-after-deluge-swept-away-dozens/

Flash floods in Pakistan kill 8 and 58 are rescued after deluge swept away dozens
PESHAWAR, Pakistan (AP) — Flash floods triggered by pre-monsoon rains swept away dozens of tourists in northwest Pakistan on Friday, killing at least eight people. The nationwide death toll from rain-related incidents rose to 18 over the past 24 hours, officials said. Nearly 100 rescuers in various groups rescued a total of 58 people and […]

Trump and Bobby Brainworm are galloping down the trail toward Soviet-style repression of science.
https://en.wikipedia.org/wiki/Repression_of_science_in_the_Soviet_Union

Why do people think music needs to be saved?
#music #Musicstodon
https://muz4now.com/2014/people-thin…

Why do people think music needs to be saved?
Stan Stewart - @muz4now

UnsafeChain: Enhancing Reasoning Model Safety via Hard Cases
Raj Vardhan Tomar, Preslav Nakov, Yuxia Wang
https://arxiv.org/abs/2507.21652 https://arxiv.or…

UnsafeChain: Enhancing Reasoning Model Safety via Hard Cases
As large reasoning models (LRMs) grow more capable, chain-of-thought (CoT) reasoning introduces new safety challenges. Existing SFT-based safety alignment studies dominantly focused on filtering prompts with safe, high-quality responses, while overlooking hard prompts that always elicit harmful outputs. To fill this gap, we introduce UnsafeChain, a safety alignment dataset constructed from hard prompts with diverse sources, where unsafe completions are identified and explicitly corrected into s…

Tinder is mandating new users in California verify their profiles using facial recognition technology starting June 30 in a bid to reduce impersonation (Kerry Flynn/Axios)
https://www.axios.com/2025/06/30/tinder-face-check-safety

Exclusive: Tinder to require Face Check for new users in California
The video selfie feature aims to curb impersonation.

As ocean acidification ramps up, experts call for speedy ocean protection https://news.mongabay.com/2025/06/as-ocean-acidification-ramps-up-experts-call-for-speedy-ocean-protection/

As ocean acidification ramps up, experts call for speedy ocean protection
Ocean health is moving into a danger zone, with rampant human-caused carbon dioxide emissions having already pushed ocean acidification levels beyond safe limits in large swaths of the marine environment, according to a recent study. The new findings underline the urgent need to ramp up protection of the world’s oceans, while simultaneously slashing CO2 emissions, […]

The Problem with Safety Classification is not just the Models
Sowmya Vajjala
https://arxiv.org/abs/2507.21782 https://arxiv.org/pdf/2507.21782

The Problem with Safety Classification is not just the Models
Studying the robustness of Large Language Models (LLMs) to unsafe behaviors is an important topic of research today. Building safety classification models or guard models, which are fine-tuned models for input/output safety classification for LLMs, is seen as one of the solutions to address the issue. Although there is a lot of research on the safety testing of LLMs themselves, there is little research on evaluating the effectiveness of such safety classifiers or the evaluation datasets used fo…