Anthropic says it can read Claude’s ‘thoughts,’ as detailed in new research paper — models observed to have a global workspace, revealing more of what makes LLM

When looking at the J-Space after Claude received prompt-injection data as part of data acquisition, Anthropic discovered the model appeared to be aware of this deception, surfacing related words like “fake, injection, f…

Colibrì proof-of-concept gains frontier-level 1.5-TB AI model — novel approach runs on only 25GB of RAM and shows promise for local AI setups

Let’s get the elephant out of the way: Colibrì’s speed on Vincenzo’s setup is only about 0.05 to 0.1 tokens per second on average, a measure that’s unusable for practical conversation — imagine just one question taking h…

AI servers will consume more power than all conventional data center hardware combined by 2027 — global data center electricity consumption set to grow by 26% t

Hyperscalers have moved in the same direction, with Meta having signed deals for more than 6GW of nuclear power to supply its upcoming data centers, and one firm repurposing retired U.S. Navy reactors for an AI site in T…

Hidden backdoor in Tenda routers goes unpatched as company ignores warnings from cybersecurity researchers — Chinese company’s firmware allows admin access with

Pending official Tenda firmware updates, CERT/CC recommends disabling remote web management wherever possible to prevent attackers from reaching the administrative interface over the internet. The organization also advis…