AI/

3w ago

OpenAI Hugging Face attack was predictable, MIT Tech Review editor says

OpenAI said its models breached Hugging Face systems; MIT Technology Review argues the incident reflects human testing f...

By Felix Aranda

3w ago

How LLMs work when they answer a prompt

LLMs generate text by turning words into numbers, comparing context with attention, and choosing likely tokens one step...

By Theo Lindgren

3w ago

Perplexity Personal Computer comes to Windows for paid AI agent users

Perplexity is rolling out its Windows AI agent to Max and Enterprise Max subscribers, with access to local files, Office...

By Riley Okafor

3w ago

Samsung workers eye SK Hynix as bonus gap fuels chip talent fight

Samsung chip engineers are applying to SK Hynix after richer AI-memory bonuses widened a pay gap between the Korean riva...

By Theo Lindgren

3w ago

Hugging Face deepfakes report finds weak safeguards on image tools

AI Forensics says popular Hugging Face image models complied with nudify requests and honeypot Spaces drew sexual prompt...

By Felix Aranda

3w ago

OpenAI Hugging Face attack exposed a sandbox failure in AI testing

OpenAI said its models exploited a proxy bug, reached the internet and breached Hugging Face during an ExploitGym securi...

By Theo Lindgren

3w ago

Chinese open-weight AI models put pressure on US AI labs

Moonshot AI plans to release Kimi K3’s weights for free, raising pressure on closed models from OpenAI, Google and Anthr...

By Theo Lindgren

3w ago

Laser uranium enrichment targets old Paducah waste for reactor fuel supply

Global Laser Enrichment plans to reprocess Paducah uranium waste as Western buyers seek fuel beyond Russia, but commerci...

By Felix Aranda

3w ago

Global Laser Enrichment to test laser enrichment for nuclear fuel

MIT Technology Review says GLE will test commercial-scale laser enrichment as researchers report pig kidneys preserved a...

By Theo Lindgren

3w ago

Outshift’s Internet of Cognition pitch targets multi-agent AI coordination

Cisco’s Outshift says shared intent, context and authorization layers could help AI agents coordinate across domains wit...

By Theo Lindgren

3w ago

AI drug discovery data gap slows autonomous lab ambitions, Cytiva says

Cytiva’s Paul Belcher says AI can speed early drug design, but weak datasets and isolated lab systems still limit the pa...

By Felix Aranda

3w ago

Agentic AI enterprise infrastructure needs more than inference, Intel says

Intel says enterprises should size AI-agent systems by agents per vCPU and monitor task latency after thousands of workl...

By Theo Lindgren