Figma is developing an in-house editing agent trained via supervised fine-tuning to understand its native file format and make iterative design changes, such as spacing, color, or layout edits, directly on the canvas. The company built training data by pairing starting designs with synthetic edit requests and validated outputs, using larger teacher models, deterministic validators, and visual judges to check results. Figma reports that an early Llama 3.3 70B model trained with this approach performed well on these tasks.
figma.com
· 2026-11-01
A developer fine-tuned the open-source Qwen3-1.7B language model on a single 16GB consumer GPU so it could respond to everyday prompts while avoiding the letter 'e' entirely, a style known as a lipogram. The project, documented on GitHub, shows before-and-after examples including a poem about trees that successfully omits 'e' after training. The creator noted the task was difficult because language models process text in tokens rather than letters, making it hard for them to track which characters appear in a word.
kevin-wu.ca
· 2026-10-25
An independent developer released qwen3-4b-deslop, a fine-tuned version of Qwen3-4B-Instruct trained on a MacBook Air using 161 example answers to remove stylistic habits like em dashes, 'not X, but Y' constructions, and lists of three. The model is available as a 2.5GB download runnable via Ollama, and on 40 held-out test prompts its measured 'slop score' fell from 121 to 12.5 per 1000 words while maintaining accuracy on 30 factual questions.
100toolkit.com
· 2026-10-19
Double Fine Productions has launched Thank You Bus Driver, a $5 chaotic driving game and the debut release from its new experimental label, Double Fine Action Labs. Players enforce increasingly absurd passenger rules while racing a bus against the clock, available on Steam, Epic Games Store, Itch.io, GOG, Humble Store, and Game Jolt. Separately, One More Level has released Valor Mortis, a $40 first-person Soulslike set during the Napoleonic Wars, starring a corrupted Napoleon voiced by Vincent Cassel, available on Steam, Xbox Series X/S, PS5, and Game Pass.
engadget.com
· 2026-10-17
Physical Intelligence has launched a beta API that lets developers upload data, fine-tune the company's π0 and π0.5 robotic foundation models on managed compute, and access the resulting models through a hosted gateway for inference. The company says the system removes the need for developers to build their own ML infrastructure or data pipelines when adapting general-purpose robot models to new tasks, environments, or hardware. It is testing the API with partner companies across industries to gauge how broadly the models generalize.
pi.website
· 2026-10-13
The US Department of Justice, working with the State Department, has filed to intervene in X's legal challenge against a €120 million EU fine at the General Court in Luxembourg. The penalty was issued in December after a two-year EU investigation concluded X misled users by labeling paid blue-checkmark accounts as 'verified.' The court will decide whether the US government may join the case.
wired.com
· 2026-09-25
The US Department of Justice has applied to the EU's General Court to intervene in Elon Musk's and X's legal challenge against a €120m EU fine issued in December 2025 over the platform's blue tick verification badges. The DOJ argues the European Commission overstepped its jurisdiction in applying the fine to an American company. The European Commission responded that it has a strong case based on X's alleged breach of the Digital Services Act.
bbc.co.uk
· 2026-09-25
404 Media reports that multiple contractors hired to improve OpenAI's models were terminated after being caught using AI tools to complete their training tasks, citing internal documents and interviews with contractors. One worker shared a termination letter citing concerns about the 'authenticity' of their work, while sources said such firings are common across a workforce of thousands. Two of the contractors worked through Mercor, a staffing firm reportedly in talks with Nvidia for a $20 billion valuation.
slashdot.org
· 2026-09-24
Ireland's Data Protection Commission, acting for the EU, fined Google €403 million (about $462 million) for mishandling location history and web/app activity data. The regulator found Google failed to be transparent about Location Accuracy on Android and unlawfully retained users' location data longer than necessary.
androidauthority.com
· 2026-09-22
Ireland's Data Protection Commission concluded a five-year inquiry into how Google Ireland handled location data across Web & App Activity, Location History and Location Accuracy features between 2018 and 2020. The regulator found multiple GDPR violations spanning lawfulness, transparency, accountability and data retention, and has fined the company €403 million while giving it six months to fix its practices.
dataprotection.ie
· 2026-09-21
Regulators concluded that Google mishandled location tracking, gathering and storing data that could expose sensitive personal details about users without adequate safeguards. As a result, the company has been fined $462 million.
fastcompany.com
· 2026-09-21
An independent researcher used supervised fine-tuning and agentic reinforcement learning to train a small 4B-parameter open-weights language model to generate PostgreSQL execution plans. On a set of 113 join-heavy queries, the tuned model cut latency by 44.7% on average compared to Postgres's default planner, even though the base model initially failed to produce valid plans for 99 of those queries. The project also involved building a custom measurement rig to reduce caching noise and a modified GRPO reinforcement learning method for scoring plans in a noisy environment.
rohanbansal.com
· 2026-09-16
GoKawiil