Hugging Face Releases DoctoBERT for French Medical Text
Hugging Face introduced DoctoBERT, a pretrained medical encoder, achieving state-of-the-art results on the DrBenchmark and a real-world clinical NER task.
Archive
Browse all published articles.
ARCHIVE
-
-
Meta Disables Muse Image Feature That Generated AI Photos of Instagram Users
Meta removed a controversial Muse Image feature that allowed AI-generated photos of Instagram users without consent after backlash. The feature was active by default and required manual opt-out.
-
S&P Global Cuts Oracle's Credit Rating Over OpenAI Risk
S&P Global downgraded Oracle's credit rating to BBB- due to OpenAI's financial risks, citing potential losses from the AI firm's collapse.
-
Brown Professor Finds AI Cheating Linked to Sharp Drop in Exam Scores
Brown University professor notes grades fell from 96 to 48 percent after switching from take-home to in-person exams, raising concerns about AI use in education.
-
OpenAI CEO Altman Says AI Is Net Job-Creating
OpenAI CEO Sam Altman now claims AI is net job-creating, a shift from earlier warnings of mass layoffs.
-
Claude Cowork's Biggest Use Case Is Mundane Office Work, Anthropic Says
Anthropic reports half of Claude Cowork usage is for administrative and text-based tasks, with business operations accounting for 33.4% of sessions.
-
Quantum Computing Boosts AI Drug Discovery Models
A Danish team used quantum computing to enhance AI models, generating more effective peptides for vaccine development.
-
VKUE Runs 34.7B Model on CPU Without GPU
Hugging Face's VKUE engine runs a 34.7B parameter model on an 8 GB laptop and a GPU-less CPU server, achieving 20 tok/s and 17 tok/s respectively.
-
Terrorist Groups Use Major AI Chatbots for Attack Planning
A new study reveals that terrorist groups are using major AI chatbots like ChatGPT and Claude for attack planning and weapons development, with Boko Haram training commanders to bypass AI safety filters since 2023.
-
OpenAI Admits ChatGPT Work Launch Had Issues
OpenAI faced backlash after launching ChatGPT Work and GPT-5.6 Sol, with users reporting exhausted limits and confusing app changes.
-
Meta's Muse Spark 1.1 Outperforms GLM-5.2 in Coding
Meta's Muse Spark 1.1 scores 71.3 on the Coding Index, outperforming GLM-5.2's 68.8 and costing $0.26 per task.
-
BAAI's Orca World Model Matches Robotics Systems Without Action Labels
BAAI's Orca world model performs as well as specialized robotics systems across five tasks, despite being trained without action labels. The model uses 125,000 hours of video and 160 million event descriptions for training.