← News & Analysis
Archive · 3349 stories

News archive.

Earlier AI news and analysis, kept for reference. Current coverage lives in News & Analysis.

Verified
May 8, 2026 · 2 min

Top AI talent avoids pharma careers despite industry demand

MIT and Harvard researchers like Mazdak Abulnaga represent the AI expertise drug companies need but struggle to recruit.

Verified
May 8, 2026 · 2 min

FDA delays Leqembi subcutaneous approval to August 24

Biogen and Eisai face a two-month delay as the FDA requests additional information on their subcutaneous Alzheimer's drug formulation.

Verified
May 8, 2026 · 2 min

AI fabricated citations jumped sixfold in academic papers

Columbia study finds fake citations rose from 1 in 2,828 papers in 2023 to 1 in 458 in 2025, with AI chatbots blamed for the pollution of scientific literature.

Verified
May 8, 2026 · 2 min

xAI seeks Anthropic partnership amid competitive pressure

Musk's AI company explores deal with Claude maker, suggesting xAI trails in foundation model competition despite massive infrastructure spend.

Incremental
May 8, 2026 · 2 min

NVIDIA releases Model Optimizer for FP8 quantization

New library handles post-training quantization to FP8 format, with CLIP demo showing minimal accuracy loss on three benchmarks.

Verified
May 8, 2026 · 2 min

GB200 clusters need block scheduling to avoid 90% bandwidth drops

NVIDIA's 72-GPU racks deliver 1.8TB/s per GPU internally but drop to 50GB/s across boundaries, forcing new Slurm scheduler constraints.

Verified
May 8, 2026 · 2 min

AMD ROCm trains medical AI model in 5 minutes, no CUDA needed

HuggingFace ecosystem runs medical question-answering fine-tuning on AMD MI300X hardware with three environment variables.

Incremental
May 8, 2026 · 2 min

Simplex cuts web app development time 40-70% using OpenAI agents

Japanese consulting firm measures significant time savings across design, development, and testing phases using ChatGPT Enterprise and Codex.

Verified
May 8, 2026 · 2 min

ChatGPT adds crisis contact alerts for self-harm detection

OpenAI's Trusted Contact feature notifies designated contacts when automated systems detect serious self-harm discussions in ChatGPT conversations.

Verified
May 8, 2026 · 2 min

OpenAI launches ads in ChatGPT for free users in five countries

Free and Go tier users in the US, Canada, Australia, New Zealand, and UK will see sponsored content while paid tiers remain ad-free.

Incremental
May 8, 2026 · 2 min

OpenAI ships three voice models with reasoning and live translation

GPT-Realtime-2 adds reasoning to voice interactions while new models handle live translation across 70+ languages and streaming transcription.

Verified
May 8, 2026 · 2 min

OpenAI launches GPT-5.5-Cyber for critical infrastructure defense

OpenAI expands Trusted Access for Cyber with GPT-5.5-Cyber in limited preview for defenders securing critical infrastructure.