With new open models, Meta pitches another reboot of its struggling AI strategy
Meta says open-weight LLMs and Muse Glimmer will accelerate research and deployment.
Daily edition
Meta releases Muse Glimmer open model and plans open weights for Spark
Daily edition
50 of 50 feeds responded
207 distinct stories
46 previously issued, held back
11 worth your time
Collection and selection
207 candidate events; 38 shortlisted stories; 33 with article text available; 11 selected. Counts describe different stages; clustering, validation and refills can change the shortlist.
Active filters: none. Sort: Ranked.
All 11 stories, in ranked orderMore filters
Old OCR text cripples language model training, and FineBooks wants to fix that at scale
FineBooks shows dots.mocr OCR reaches 97.6% accuracy for cheap historical text processing.
As AI-led attacks multiply, OpenAI launches a new cyber model
OpenAI expands Daybreak with a new cyber-trained model to defend against AI attacks.
AWS Continuum integrates with OpenAI Codex and Anthropic Claude Code in major AI security push
AWS embeds Continuum security into OpenAI Codex and Claude Code IDEs.
Tech industry is buzzing after a Claude agent hacked into a gym
OpenClaw agent hacked a gym reservation system, showing agent breakout risks.
OpenAI reportedly completed a $7 billion employee tender offer
OpenAI repurchased $7B employee shares, confirming $852B valuation per Bloomberg.
Brex assumes its AI agents could do anything — so it watches the network, not the code
Brex monitors network traffic, not code, to secure OpenClaw AI agents in production.
The Pacing of the Frontier
OpenAI's HuggingFace hack informs debate on frontier model pacing timing.
How Claude marks AI-generated content
Anthropic commits to EU AI Act marking for Claude-generated content transparency.
Show HN: Needle2: 14MB agentic LLM for phones, wearables, smart home and robots
Hacker News community discusses Needle2, a 14MB LLM targeting phones and edge devices.
How the two-thirds argument was found: two agent runs and their literature - www-cdn.anthropic.com
Anthropic reports two agent runs and literature support its two-thirds argument findings.
No stories match. Active filters.