New AI Safety Method HardFlow Developed by MIT
MIT researchers have introduced HardFlow, a new technique that enables generative AI models to meet strict safety requirements without needing retraining.
Latest reporting
37–48 of 49 stories
Finland’s Largest Viking Age Silver Hoard Found
A metal detectorist in Sysmä, Finland, has unearthed the country's largest hoard of Viking Age silver, weighing ten pounds and dating to the 11th century.
FBI Probes Massive Driver’s License Data Breach
The FBI is investigating a dark web service called Nexus that claims to hold over 153 million driver's license records from the U.S. and Canada.
NASA Launches Artifact InSPECtor Citizen Science Project
NASA has introduced Artifact InSPECtor, a new citizen science initiative that invites volunteers to help train artificial intelligence models to identify and remove data errors…
Nvidia Revenue Growth Forecasts 70% Next Year
CEO Jensen Huang projects Nvidia revenue growth of 70% next year, citing the company's foundational role in the global AI ecosystem and strong contract demand.
AI Agents Flooding Public Services With New Requests
Public services globally are seeing a surge in applications as AI agents flooding systems help citizens navigate complex administrative burdens.
Hackers Are Stealing Claude Tokens From Subscribers
Anthropic has confirmed that hackers are using infostealer malware to compromise Claude login sessions and consume paid subscriber tokens.
Roblox Releases New AI Child-Safety Tools
Roblox has shared updated AI safety models with the ROOST community to help platforms better detect online grooming and personal information requests.
Authors Dispute Anthropic Settlement Payment Claims
Authors are challenging claims made by publishers and agents on their Anthropic settlement payments, citing widespread record-keeping errors.
Woman Joins xAI Grok Lawsuit Over Explicit Images
A woman has joined an xAI Grok lawsuit, alleging her stepfather used the chatbot to create thousands of explicit images from a childhood photo.
Gene.01 Humanoid Robot Uses Smart Skin for Safety
The new Gene.01 humanoid robot features sensor-filled skin to detect workers before contact, aiming to improve industrial safety through Physical AI.
YouTube Monetization Policy Update for AI Content
YouTube has updated its monetization policy to restrict AI-generated inauthentic content, including repetitive videos and AI personas discussing sensitive topics.
