Technology

New Mexico hails ‘historic’ win after jury finds firm misled consumers over safety and enabled harm against users

Ngl, I wouldn't even be able to turn down that offer lmao

Social media platforms Instagram and YouTube have a design defect which means they are addictive, a jury in the United States has ruled.

The Los Angeles jury took nearly nine days to reach its verdict in the landmark case brought by a woman known as KGM against social media platforms. It awarded US$3 million in damages, with Meta (owner of Instagram) being 70% responsible and Google (owner of YouTube) 30%. The jury later awarded a further US$3 million in punitive damages.

Both TikTok and Snap settled on confidential terms before the six-week trial commenced.

This is Meta’s second big loss in the US courts this week, with a New Mexico jury finding the company guilty on March 24 of concealing information about the risks of child sexual exploitation and the harmful effects of its platforms on children’s mental health.

https://archive.ph/20260326162202/https://www.404media.co/wikipedia-bans-ai-generated-content/

TurboQuant looks like a pretty massive deal for running local models efficiently. The core issue they are tackling is the memory bottleneck caused by the key value cache during generation. When you are doing long context inference storing all those high dimensional vectors eats up VRAM extremely fast. Traditional vector quantization helps but usually introduces memory overhead because you have to store scaling factors or constants in full precision for every small block of data. That overhead can easily add an extra bit or two per parameter which ruins the compression targets people are aiming for.

TurboQuant solves the problem by combining two clever mathematical tricks to eliminate that overhead entirely and get the cache down to 3 bits without losing accuracy. The first part is an algorithm called PolarQuant. Instead of looking at the vectors in standard cartesian coordinates it converts them into polar coordinates. This basically separates the magnitude from the direction. Because the angles map onto a fixed predictable circular grid the model no longer needs to store those dynamic bounding boxes or normalization constants that traditional methods require. That step handles the bulk of the compression to capture the main signal of the vector.

The second piece of the puzzle is where they use something called Quantized Johnson Lindenstrauss or QJL to clean up the residual error left over from the first step. QJL uses a mathematical transform to shrink that leftover error down to just a single sign bit of positive or negative one while preserving the relative distances between the data points. This acts as a mathematical error checker that fixes any bias in the attention scores. Because it only uses one bit and preserves the geometry of the space the attention mechanism can still calculate accurate logits without needing full precision data.

They tested this on open weights models like Gemma and Mistral across heavy needle in a haystack and LongBench tasks. They managed to compress the KV cache down to 3 bits with literally zero drop in accuracy and they did not even need to do any fine tuning or calibration. On top of saving a massive amount of VRAM the 4 bit version actually speeds up attention logit computation by up to 8x on H100 GPUs compared to standard 32 bit floats. This seems like a massive leap forward for anyone trying to run long context models on constrained hardware or scale up huge vector search databases.

LOS ANGELES — A jury found Meta and YouTube negligent in the design or operation of their social media platforms, producing a bellwether verdict in the first lawsuit to take tech giants to trial for social media addiction.

The jury stated that Meta's and YouTube's negligence were a substantial factor in causing harm to the plaintiff, identified in court by her initials, K.G.M., and that the companies failed to adequately warn users of the dangers of Instagram (Meta's platform) and YouTube (which is owned by Google).

They awarded K.G.M. $3 million in compensatory damages, finding Meta 70% responsible for harm caused to the now 20-year-old plaintiff, and YouTube responsible for 30%.

Slovenia's Foreign Minister Tanja Fajon said she regretted the government's move not to join South Africa's genocide case against Israel at the ICJ, claiming that external "pressure" had contributed to the decision.

Slovenia decided against participating due to "security risks". While Prime Minister Robert Golob had initially been inclined to give the proposal the green light, he was ultimately swayed against doing so by national security officials, local media reported.

They reportedly cautioned that joining the lawsuit could jeopardise Slovenia's national security, noting that many of the country's cyber defence systems are of Israeli origin.

• Invidious
• Youtube

https://archive.ph/2026.03.24-212135/https://www.bloomberg.com/news/articles/2026-03-24/openai-plans-to-discontinue-support-for-sora-ai-video-generator

https://archive.ph/2026.03.24-233256/https://www.bloomberg.com/news/articles/2026-03-24/arm-to-sell-its-own-chips-for-first-time-in-bid-for-ai-revenue

• Invidious
• Youtube