AI News for May 4, 2026 — 7min.ai

[RESEARCH] MIT explains why scaling laws work: superposition packs concepts into too few dimensions: MIT researchers Yizhou Liu, Ziming Liu, and Jeff Gore traced the reliability of LLM scaling laws to a geometric property called superposition, where models stash tens of thousands of concepts into a few thousand dimensions by letting their vectors overlap.
[NEWS] VS Code developer confirms Copilot attribution tag was unauthorized, revert coming in 1.119: A Microsoft developer behind the controversial pull request acknowledged that silently adding 'Co-Authored-By: GitHub Copilot' to all commits was a mistake, and said the default will be reverted in VS Code version 1.119.
[NEWS] Hangzhou court orders compensation after QA engineer refused AI-driven 40% pay cut: The Hangzhou Intermediate People's Court ruled that a Chinese tech company illegally terminated a quality-assurance engineer after an LLM took over his task-checking role.
[RESEARCH] CAISI: DeepSeek V4 Pro lands eight months behind US frontier models: The US Center for AI Standards and Innovation tested DeepSeek V4 Pro across cybersecurity, software development, math, natural sciences, and abstract reasoning, concluding the open-weight Chinese flagship sits roughly eight months behind leaders like Opus 4.6 and GPT-5.4.
[NEWS] Kenya's AI-driven health means-testing systematically overcharges the poorest: An investigation by Africa Uncensored, Lighthouse Reports, and the Guardian audited the predictive ML algorithm Kenya uses to set healthcare contributions under President Ruto's Social Health Authority, launched October 2024.
[NEWS] Altman's trial lawyer: Musk's real goal is to weaken a rival to xAI, not protect a charity: Altman's lead attorney, who previously defeated Musk in the 2022 Twitter acquisition lawsuit, told the Oakland jury that Musk's nonprofit-betrayal claim is driven by competitive motives, arguing his real aim is to damage a rival to his own xAI.
[NEWS] Sam Altman: companies are 'AI washing' layoffs they were going to do anyway: OpenAI CEO Sam Altman told CNBC-TV18 at the India AI Impact Summit that some employers are blaming AI for workforce cuts they had already planned, alongside genuine displacement.
[OPINION] AI music hits 34% of Deezer uploads as 50,000 tracks land daily: Deezer reported fully AI-generated music accounted for 34% of new uploads by late 2025, climbing from 28% in September and now exceeding 50,000 tracks per day.
[NEWS] UK biometrics watchdogs: facial recognition oversight is years behind the rollout: England and Wales biometrics commissioner William Webster and Scotland counterpart Brian Plastow told the Guardian that legislation is failing to keep pace with police and retailer use of live facial recognition.
[NEWS] ASX warns listed companies against 'ramping' stock prices with inflated AI claims: Australia's stock exchange operator ASX told listed companies not to overstate AI's impact on their operations, saying it actively monitors disclosures for so-called ramping that pushes share prices on weak substance.

About 7min.ai · Subscribe to newsletter