ENFR
8news

Tech • IA • Crypto

TodayTopicsVideosCryptoArchivesFavorites

IPTV Botnet Backdoors, GPT-5.6 vs Gemini 3.5, China AI Curbs

AITuesday, July 7, 2026· 12 videos

Briefing

Audio player
0:00 / 0:00

IPTV boxes expose global botnet

Millions of illicit IPTV devices have been found operating as remotely controllable nodes in a global botnet. Analysis revealed constant network probing, ARP scans, and connections to services like Tencent QQ, indicating coordinated activity. Devices ship with ADB enabled without authentication and even TeamViewer preinstalled, allowing silent remote access. Built on outdated Android Open Source Project 2021 firmware, they present systemic vulnerabilities with potential use in cybercrime or espionage.

UN launches AI governance commission

A United Nations-backed “AI for Good” commission has been formed in Geneva, bringing together 44 global leaders. Co-chaired by Rwanda’s president and Salesforce CEO Marc Benioff, it includes firms like Microsoft and Nvidia. The group focuses on trust, access, and impact, aiming to shape long-term coordination rather than enforce regulation. Early reports warn of inequality, economic disruption, and the need for shared global infrastructure.

GPT-5.6 vs Gemini 3.5 race

A tightly clustered release window is intensifying competition, with GPT-5.6 expected between July 7–9 and Gemini 3.5 Pro alongside DeepSeek V4 on July 17. Google DeepMind reportedly restarted pre-training, signaling a deeper architectural shift rather than incremental updates. The focus is shifting from scale to reasoning quality, code generation, and efficiency. This marks one of the most concentrated bursts of frontier model launches to date.

China weighs AI export restrictions

China is considering limits on exporting advanced AI models, consulting firms like Alibaba, ByteDance, and Z.ai. Proposals include mandatory reviews or outright bans on releasing frontier systems abroad. The move mirrors U.S. restrictions on models such as Fable 5 and Mythos 5, signaling a global shift toward AI sovereignty. Reduced access to models like Qwen and GLM 5.2 could raise costs and fragment global AI development.

Microsoft Xbox cuts 3,200 jobs

Microsoft’s Xbox division will eliminate 3,200 roles by 2027, with 1,600 immediate layoffs. The cuts affect roughly 20% of the workforce and follow persistent profitability issues. Internal data shows the unit loses about $0.64 per dollar invested and suffers margins far below competitors. Microsoft plans to streamline operations, reduce management layers, and spin out studios like Ninja Theory and Arkane.

Distillation revives local AI models

Developers are recreating advanced model capabilities using distillation techniques and large-scale usage traces. Datasets of around 2.3 million examples capture reasoning, coding, and tool-use workflows. Resulting models range from 4B to 30B parameters, small enough for local deployment yet capable of advanced tasks. Techniques like quantization (Q2–Q8) balance performance and efficiency, reducing reliance on costly proprietary systems.

Sephora scales AI across retail

Sephora is deploying AI across operations, reaching 37 countries, 3,000 stores, and 80 million loyalty members. A conversational assistant and ChatGPT integrations are driving over 5% higher online conversion rates. AI supports personalization, marketing optimization, and supply chain forecasting at scale. The strategy positions AI as core infrastructure for customer experience and revenue growth.

AI accelerates digital twin medicine

Research led by Derya Unutmaz highlights AI’s growing role in building personalized digital twins of human biology. Advances in reasoning models (2024) enable hypothesis generation and complex immunology insights. Tools like Codex dramatically reduce development time for scientific software, from months to days. The approach could transform medicine by enabling simulation-driven, individualized treatment strategies.

Videos covered

Previous briefings · AI