Urgent.News

What's breaking now, across thousands of outlets.

Thursday, 5 November 2026 Reading Room

Tech

SilentMind: AI that censors and generates controversy in searches

SilentMind: the censor model that is booming in searches and generating controversy Introduction On October 15, 2024, Tencent launched SilentMind, an AI model that automatically blocks sensitive and political topics. The difference? When the filter is triggered, the response is cut off, replaced by “content not available” or generic messages. The announcement triggered Google Trends (↑ 420% in…

Translated from Portuguese Read in Portuguese

dev.to ·

Tencent launched SilentMind, an AI model that automatically blocks sensitive and political topics, on October 15, 2024. The model cuts off responses and replaces them with generic messages or "conteúdo não disponível" when it detects sensitive keywords. This has sparked a debate about freedom of expression, competition, and compliance, with Google Trends showing a 420% increase in searches related to "SilentMind censura" in Brazil, Portugal, and Mexico.

The article analyzes SilentMind's architecture, compares its performance with other AI models, and provides strategies for detecting and circumventing its filters.

Written by urgent.news from Dev.to's report — not a translation of it. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in Tech

Tencent's SilentMind: discover and evade its censorship filters

Tencent's SilentMind: What does it filter and how to bypass it? Introduction In less than 48 hours after the announcement, Tencent's SilentMind became the most searched topic on Google Trends. Why?

  • SilentMind AI model filters sensitive/political content
  • Detect truncated responses with Python code snippet
  • Alternative: use multiple models for less filtering

·dev.to

Keyword Research in Python: Ideas, Search Volume, Difficulty and Intent for ~$3 per 1,000 Keywords

Most keyword research comes down to one loop: start from a few seed terms, expand them into everything people actually search, then keep the ones with real volume and a difficulty you can win.

  • Python-based tool automates keyword research process
  • Provides search volume, CPC, KD, intent, SERP features, trend percentages
  • $3 per 1,000 keywords with filtered results based on various criteria

·dev.to

Cheapest per token is not cheapest per finished job: what 2,400 runs show

A model with a 33% pass rate turned out to be the cheapest way to finish a task in a recent 10-model benchmark. The same benchmark shows why that result is less of a free lunch than it looks.

  • Cheapest model gpt-oss-120b 12 times cheaper per finished task than GPT-5.5
  • Coverage crucial factor: gpt-oss-120b solved 8/40 tasks, GPT-5.5 solved 25/40
  • Price tier not always predicts success: GPT-5.5 69% pass rate on easy tasks

·dev.to

More from Thursday 5 November →