| ZOTPAPER |
| |
| News without the noise |
| Tuesday 6 October 2026 · The Daily Edition |
| |
From the editor A front page that traces the boundaries of AI, from the hardware it runs on to the safety it demands and the security it can undermine. |
Geopolitics · 2 min
Chinese chipmakers have accumulated an estimated 343 immersion DUV lithography tools from ASML since 2012, many capable of producing 7nm-class chips, according to a report from the Centre for Technology and Statecraft, a Washington-based research group. The finding underscores the limits of existing export controls and has renewed calls for stricter regulations on semiconductor equipment sales to China.
Why it leads China's stockpile of advanced chipmaking tools exposes the high-stakes race for semiconductor self-sufficiency and the limits of export controls. |
| |
AI · 3 min
OpenAI chief executive Sam Altman has said the world should accept 'bad things' happening with artificial intelligence, including hacks and scams, in exchange for the technology's benefits, sparking immediate backlash from safety advocates and political figures. In an interview released Monday, Altman contrasted OpenAI's lighter-touch regulatory stance with stricter approaches, arguing that society must tolerate some misuse to realise AI's potential.
Why it leads Sam Altman's blunt call to accept AI harms for progress forces the industry to confront a fraught ethical trade-off. |
| |
Security · 3 min
Cloudflare has patched a cross-tenant data exposure vulnerability in its Containers platform after a researcher showed that paying customers could recover residual disk blocks left behind by other customers' workloads on the same host. The company says it found no evidence the flaw was exploited maliciously.
Why it leads Cloudflare's container flaw that leaked other customers' data is a sharp reminder that infrastructure security remains a moving target. |
| |
Editor's picks AI · Safeworld's stealth exit with $12M for AI robot safety simulations signals growing demand for rigorous testing before deployment. Security · Cloudflare's plan for a quantum-safe certificate authority prepares the internet for a post-quantum world, a necessary but enormous undertaking. Security · Google's pause of its open-source bug bounty after an AI-generated report flood reveals a new vulnerability in the vulnerability discovery process. Security · The Qantas hack suspect's cooperation with the FBI adds a developing twist to one of Australia's most consequential data breaches. |
Today's segment · Paper of the Day
Neural network precision floor follows a depth law predicted by amplification
|
|
arXiv:2609.32060v1
Ahmad S. Tarawneh — Mutah University
The authors define the precision floor of a neural network as the perturbation level or bit-width at which test accuracy falls halfway from full precision to chance. They show that this floor is given by η_c = Λ / G, where G is the predictive amplification, and that G² grows linearly with depth. Across MLPs, CNNs, Vision Transformers and nine pretrained language models, the depth exponents of the floor and amplification agree within 95% intervals in twelve of thirteen trained architectures. They also report a quantization-aware training (QAT) paradox: noise-aware training roughly doubles tolerable noise in shallow networks, but the gain decays with depth, steepening the depth law.
|
|
|
| |
Unsubscribe · Manage preferences · View in browser
zot.news · Privacy · editor@zot.news
You're receiving this because you subscribed to the Zotpaper Daily at zot.news.
Zotpaper · Brisbane, Australia
|