Plants
Frontier lab releases model with longer context window
Early benchmarks show gains on long-document tasks, with mixed results on reasoning.
Plants
Frontier lab releases model with longer context window
Early benchmarks show gains on long-document tasks, with mixed results on reasoning.
Makers
Open source model closes gap on proprietary benchmarks
A community-trained release now scores within a few points of the closed-source leaders on two major evaluations.
Suppliers
Regulator proposes disclosure rules for frontier training runs
The draft rule would require labs above a compute threshold to report training runs before they begin.
Production
Two labs settle dispute over training data sourcing
Terms were not disclosed, but the case had been closely watched as a bellwether for the industry.
Policy
Enterprise software giant adds agent framework to core product
The rollout targets internal workflow automation rather than customer-facing chat.
15 stories today6 sections30 topics trackedUpdated through the day
Latest
Makers
Open source model closes gap on proprietary benchmarks
A community-trained release now scores within a few points of the closed-source leaders on two major evaluations.
Suppliers
Regulator proposes disclosure rules for frontier training runs
The draft rule would require labs above a compute threshold to report training runs before they begin.
Production
Two labs settle dispute over training data sourcing
Terms were not disclosed, but the case had been closely watched as a bellwether for the industry.
Policy
Enterprise software giant adds agent framework to core product
The rollout targets internal workflow automation rather than customer-facing chat.
News
Startup raises large round for AI coding assistant
The round values the company well above its last raise just over a year ago.
Plants
Acquisition brings voice AI startup under major platform
The deal folds a small but well regarded voice model team into a much larger product organization.
Makers
New paper questions reliability of common reasoning benchmark
The authors argue a widely cited benchmark may be more contaminated by training data than previously assumed.
Suppliers
Researchers publish method for detecting synthetic training data
The technique could help labs audit whether their training sets contain unlabelled AI generated content.
Production
Major platform ships agent mode in browser
The feature lets the assistant complete multi-step tasks across open tabs with user confirmation at each step.
Policy
Productivity suite adds meeting summary feature
The rollout follows a wave of similar features across competing platforms this year.
News
Chip maker unveils next generation AI accelerator
The new chip promises a meaningful efficiency gain over the current generation, with samples shipping to partners this quarter.
Plants
Cloud provider announces new data center region for AI workloads
The facility is purpose built for large scale training rather than general cloud compute.
Explore topics
JapanChinaIndiaSouth KoreaThailandIndonesiaToyota GroupVolkswagen GroupHyundai Motor GroupStellantisGeneral MotorsFordBoschDensoZFMagnaAisinContinentalOutput & capacityCapacity utilisationLine stoppagesModel changeoverShift patterns & labourAutomation & robotsTariffs & trade rulesLocal content requirementsEuro 7China VICAFEEV mandates
Common questions
How often is The Plant Floor updated?
New stories are published through the day as reporting is confirmed. The homepage lists the most recent items first.
Where does The Plant Floor get its information?
Every story carries a named source attribution. Our methodology page sets out which document types we treat as primary and how we handle estimates.
Who writes the articles?
Each story carries a named byline. Author pages collect everything that writer has filed for this site.
How are corrections handled?
Corrections are logged on a dedicated page with a date and a plain description of what changed, rather than edited in silently.