Meta has introduced new AI tools powered by large language models to detect 'signposting' ads that secretly direct users to illegal child sexual abuse material. In H1 2026, its systems proactively removed 33.2 million pieces of violating content, with over 97% detected before user reports. Concurrently, WhatsApp launched enhanced parental controls for teen accounts.
Content Moderation
内容审核
The process of reviewing user-generated content to ensure it complies with community guidelines.
Curated Intelligence
OpenAI recently banned a group of likely PRC-origin accounts that used AI to generate comments and cartoons regarding US tech policy, tariffs, and trade restrictions. This action highlights the potential risks of generative AI in political influence operations and the platform's latest governance measures against abuse.
OpenAI outlines its safety governance mechanisms for extreme scenarios like mass shootings and violent threats, detailing how models distinguish between fictional/historical discussions and real-world harm. The company emphasizes balancing privacy with law enforcement collaboration, incorporating insights from psychologists and legal experts to refine risk detection.
OpenAI has released prompt-based teen safety policies for developers using the gpt-oss-safeguard model. This initiative aims to help AI systems automatically identify and mitigate age-specific risks through standardized prompt engineering techniques.
Google has released the preview version of Gemini 3.1 Flash-Lite, designed specifically for high-volume workloads. Priced at $0.25 per million input tokens and $1.50 per million output tokens, it claims superior cost-efficiency and speed compared to the previous 2.5 Flash model, targeting tasks like translation, content moderation, and UI generation.
OpenAI banned an account linked to a Chinese law enforcement individual for allegedly using AI to plan influence operations, harassment, and cyber activities. This highlights the potential misuse of generative AI in malicious social engineering and information warfare.
OpenAI banned accounts linked to a Russia-origin influence operation dubbed “Stop News,” which used AI to generate propaganda targeting Africa and the UK. This highlights the growing risk of generative AI being weaponized for cross-border information manipulation.
OpenAI has banned several Korean-language accounts that used its models to support malware development, debugging, phishing, and credential-theft workflows. This action aims to curb the use of LLMs in facilitating cybercrime operations.
OpenAI announced that safety is foundational to the launch of its Sora 2 video model and social app. The company emphasizes concrete protections to address novel risks from state-of-the-art video generation. Current information focuses on governance principles rather than specific technical implementations or metrics.
OpenAI outlines its strategy to prevent, detect, and disrupt AI-generated child sexual abuse material (CSAM), detailing prohibitions on using models for exploitation. The announcement covers protections from pre-deployment to production and expresses support for New York state legislation ensuring legal safeguards for responsible reporting.