AI/

4d ago

Alibaba Qwen3.8-Max arrives with frontier-model claims and caveats

Alibaba says Qwen3.8-Max is its largest AI model yet, but its performance claims and release status require careful read...

By Riley Okafor

6d ago

Fenix Flexin ‘Rubberz’ AI questions follow its Hot 100 rise

The song has reached No. 58 on Billboard’s Hot 100 as listeners and The Verge point to artifacts common in AI-generated...

By Theo Lindgren

6d ago

AllFaith Benchmark ranks Aligned AI closest to neutral

University researchers tested faith-and-ethics bias in frontier AI models, with Aligned AI scoring nearest to neutral.

By Theo Lindgren

1w ago

Google Earth AI tool pulled after fake satellite imagery spreads

Google rolled back a prompt-based Earth image editor after users shared synthetic scenes that appeared to breach its rul...

By Theo Lindgren

1w ago

Google Earth AI image generator raises fake satellite image risks

Google’s new Earth feature can alter map imagery from prompts, and Henk van Ess has shown how credible-looking fakes can...

By Theo Lindgren

1w ago

Montana right to try law opens a narrow path to unapproved drugs

A Montana program could let patients buy experimental drugs, but one father’s search for treatment shows how narrow that...

By Felix Aranda

1w ago

Apple’s iCloud Plus AI tier could raise usage limits for heavy users

Tim Cook said Apple expects to offer iCloud Plus upgrades for people who use Apple Intelligence and the new Siri AI heav...

By Felix Aranda

1w ago

Montana experimental drug access rules move unapproved treatments toward clinics

Montana finalized rules letting consenting adults buy unapproved drugs through clinics, raising FDA, safety and price co...

By Theo Lindgren

1w ago

Friend AI pendant returns with a speaker and a $249 price

Friend’s relaunched AI pendant can now talk aloud through a speaker, raising the price from $129 to $249 after a noisy f...

By Felix Aranda

1w ago

Gemini Robotics 2 gives Google DeepMind’s robots full-body control

Google DeepMind says Gemini Robotics 2 can coordinate humanoid legs, arms and hands, moving its demos beyond tabletop ma...

By Riley Okafor

1w ago

LLM fundamental flaw leaves popular models open to jailbreaks, researchers say

Researchers say a basic instruction-handling problem lets popular LLMs reveal blocked information and may resist a compl...

By Felix Aranda

1w ago

LLM security flaw lets attackers spoof model roles, researchers say

Researchers say models infer instruction roles from writing style, letting attackers bypass safeguards in tests across m...

By Riley Okafor