Folarin Balogun is an American
and what Mike Johnson continues to make clear is that
he despises Americans.
https://bsky.app/profile/atrupar.com/post/3mp26lz4tis2f
At this small scale, the attribution to #OpenStreetMap is unnecessary. But because people often omit the credit, it’s actually refreshing to see this proper use of OSM data by #DWNews (via #MapTiler).
V…
Ho pubblicato una traduzione veloce di "Valery" di #AlfredoCohen / #FrancoBattiato. Sono in difficoltŠ con un paio di versi, una seconda occhiata mi aiuterebbe.
CNN blames a "technical issue" at WBD for its channels going dark on Monday across multiple distribution points (Todd Spangler/Variety)
https://variety.com/2026/tv/news/cnn-streaming-technical-issues-channel-loss-1236841649/
…
EOLE Evento europeo sul diritto all'open source e al software libero. Workshop di apertura online domani 25 giugno 2026 (dalle 15:00 alle 17:30 CEST)
L'Europa ha riscoperto la "sovranitŠ" attraverso un'ondata di nuove regolamentazioni - Legge sulla resilienza informatica, Legge sull'IA, gare d'appalto per la sovranitŠ del cloud. EOLE 2026 adotta un punto di partenza opposto: il Software Libero, l'Open Source e l'IA aperta perseguono l'autonomia…
The thing about the "Iran hacks UK power station" story is that the details are sparse and the public attribution seemingly weak.
It's not like we haven't gone through years of UK hacks being blamed on nation states and then later revealed to have been a teenager in Cardiff or whatever.
Which Model Is Actually Serving You? IRIS: Budgeted Black-Box Auditing of Model Substitution and Routing Dilution in LLM Gateways
Yuewei Zhang, Zhi-Hai Zhang, Hanzhang Qin
https://arxiv.org/abs/2607.20860 https://arxiv.org/pdf/2607.20860 https://arxiv.org/html/2607.20860
arXiv:2607.20860v1 Announce Type: new
Abstract: Commercial LLM gateways mediate access to hosted models, but the served backend may not match the advertised one: it may substitute a cheaper model on every request or route only a fraction $\epsilon$ of requests to it. Prior black-box auditors often need a privileged signal (log-probabilities, token ranks, or reference samples) or a target-specific probe, fix the query budget in advance, and return a yes/no verdict. We present $\mathrm{IRIS}$, an audit that needs only the returned text: it asks endpoints to generate random numbers or strings, fingerprints the backend, and is the first to combine, in one text-only audit, detection of whole-stream substitution and fractional dilution, attribution of the served backend, routing-fraction ($\epsilon$) estimation, and a query budget it sizes itself. A cheap pilot fits the exponential query-error decay and freezes that budget before any suspect query is issued. On an intra-family Qwen3 ladder $\mathrm{IRIS}$ verifies the backend at $0.99$ AUROC and sharpens attribution as queries accumulate; across a commercial OpenRouter library it catches $\epsilon{=}0.3$ dilution on margin-qualified pairs at $0.85$ mean power ($0.017$ false-positive rate) and recovers $\epsilon$ to within $0.04$ for enrolled diluents; and a live cross-provider audit flags $14$ of $15$ same-model provider pairs by genuine quantization and kernel deviations, corroborated on third-party MET traces. Against comparable black-box auditors, $\mathrm{IRIS}$ matches or beats detection on shared tasks, and adaptive allocation lifts the matched-budget target-hit rate from $73$% to $87$%. Further experiments cover adversarial gateways, knob identifiability, unseen diluents, and false-positive control.
toXiv_bot_toot
november17: November17 members (2009)
A network representing connections among members of the November 17 (N17) Greek terrorist group. Nodes are members, and an edge exists if two members have some connection in the past. Metadata include role, function, resources. Some attributes are missing.
This network has 22 nodes and 66 edges.
Tags: Social, Offline, Unweighted, Metadata
It's pretty clear that Trump is bisexual.
https://bsky.app/profile/atrupar.com/post/3moia4oenbb2x