LLM Benchmarking, Reimagined: Put Human Judgment Back In
If you only look at automated scores, most LLMs seem great—until they write something subtly wrong, risky, or off-tone. That’s the gap between what static benchmarks measure and what your…
2025
India’s gig workers win legal status, but access to social security remains elusive
India is restructuring its gig economy with a new labor law, but much more is needed before gig workers see real benefits.
Microsoft AI Releases Fara-7B: An Efficient Agentic Model for Computer Use
How do we safely let an AI agent handle real web tasks like booking, searching, and form filling directly on our own devices without sending everything to the cloud? Microsoft…
Google teams up with Accel to hunt for India’s next AI breakouts
Google and Accel will jointly invest up to $2 million in each startup through their new partnership.
Altman describes OpenAI’s forthcoming AI device as more peaceful and calm than the iPhone
Altman and Ive tease a simple AI device aimed at calm, distraction-free computing, launching within two years.
OpenAI learned the hard way that Cameo trademarked the word ‘cameo’
OpenAI might have to rename the “cameo” feature in the Sora app.
Rad Power Bikes’ batteries receive major fire risk warning
The U.S Consumer Product Safety Commission claims Rad Power “refused to agree to an acceptable recall.” Rad Power says the CPSC’s solution would bankrupt the company.
Tesla FSD software may not be approved by EU regulator after all
Tesla claimed in a weekend social media post that a Dutch regulator was set to approve its Full Self-Driving mode. It seems the regulator isn’t quite in line with Tesla.
Hands on with Stickerbox, the AI-powered sticker maker for kids
Stickerbox turns kids’ ideas into printable stickers, blending AI magic with hands-on coloring for a surprisingly creative, “screen-light” play experience.
AWS is spending $50B build AI infrastructure for the US government
AWS has been working with the U.S. government since 2011 and is now building AI infrastructure specifically for the entity.