Nathan Lambert: cyber risk discourse on open models is broken
Read original source →interconnects.ai
Interconnects essay argues policy debates over open-weight cyber risk assume banning open models works in a silo and that China ignores AI safety. It cites Anthropic's report on GLM-5.3 as an offensive cyber tool as the latest example of the narrow framing.
2 sources
More stories today
Neostellar principal: AI adoption will be 'unseen' in prior tech cycles
Bloomberg Intelligence says US AI labs' performance lead over China narrowed to a record low as labs like DeepSeek gained ground. Neostellar principal Willy Lee gives a VC view on AI adoption.
Bloomberg Technology·2 hours ago
Perplexity's Computer plays StarCraft against itself in real time
Aravind Srinivas·2 hours ago
AI data center developer offers Pennsylvania town $10,000 per household
Kashmir Hill·2 hours agoPinterest's AI turns beauty Pins into salon action plans
Pinterest's Beauty Guides, powered by its Pinterest Intelligence visual and generative AI, translate hair and nail Pins into stylist terms like "balayage" and "root melt," with cost estimates, appointment times and maintenance needs. At launch it covers hairstyle, color, nail shape and finish Pins.
TechCrunch·2 hours ago
Google picks 100+ startups for fourth Gemini Startup Forum
Founders from more than 100 startups chosen from over 2,000 applications will meet at Google's Mountain View HQ in November for a two-day summit. Participants represent 17 countries and get up to $350,000 in Google Cloud credits.
Google Blog·2 hours ago
Researcher argues AI field was "born toxic" from automata studies origins
Margaret Mitchell·2 hours agoFDA outlines 2027 plans for AI guidance
STAT's Health Tech newsletter reports the FDA has spelled out its 2027 AI guidance plans. The item is subscriber-only, with the full article behind a STAT+ paywall.
STAT News AI·2 hours ago
Rippling AI runs on LangChain Deep Agents and LangSmith
LangChain·2 hours ago
