PulseAugur
EN
LIVE 20:05:44

AI processor boosts efficiency; AI tool security risks emerge

A new photonic processor from Stuttgart-based Q.ANT promises significant energy efficiency gains for AI and HPC workloads, performing calculations natively in the optical domain. This co-processor, built on thin-film lithium niobate, offers up to 30x the energy efficiency of traditional chips and is already deployed in major European supercomputing centers. Separately, an AI agent on a Mastodon server detected a cryptominer disguised as a system helper, highlighting the security risks of unauthenticated AI tools. AI

IMPACT New photonic processors could drastically cut AI compute costs, while security incidents highlight the need for robust authentication of AI tools.

RANK_REASON Cluster covers both a new hardware product release for AI and a security incident involving AI tools.

Read on Mastodon — mastodon.social →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

COVERAGE [2]

  1. Mastodon — mastodon.social TIER_1 English(EN) · datarazimedia ·

    An AI agent on a live server flagged a dotfile in temp pretending to be a system helper. It was a cryptominer running as root, in a container for an AI tool pub

    An AI agent on a live server flagged a dotfile in temp pretending to be a system helper. It was a cryptominer running as root, in a container for an AI tool published online with no authentication. Worse: a backdoor key had been added to the operator's account, and someone was al…

  2. Mastodon — mastodon.social TIER_1 English(EN) · datarazimedia ·

    Stuttgart-based Q.ANT has shipped a commercial light-based processor for AI and HPC. Built on thin-film lithium niobate, it does the maths natively in the optic

    Stuttgart-based Q.ANT has shipped a commercial light-based processor for AI and HPC. Built on thin-film lithium niobate, it does the maths natively in the optical domain — so almost no on-chip heat, and up to 30x the energy efficiency of conventional chips on the same AI workload…